The use of functional placental magnetic resonance imaging for assessment of the placenta after prolonged preterm rupture of the membranes in vivo: A pilot study

Preterm prelabor rupture of membranes (PPROM) complicates 3% of pregnancies in the UK. Where delivery does not occur spontaneously, expectant management until 37 weeks of gestation is advocated, unless signs of maternal infection develop. However, clinical presentation of maternal infection can be a late sign and injurious fetal inflammatory responses may already have been activated. There is therefore a need for more sensitive markers to aid optimal timing of interventions. At present there is no non‐invasive test in clinical practice to assess for infection in the fetal compartment and definitive diagnosis of chorioamnionitis is by histological assessment of the placenta after delivery. This study presents comprehensive functional placental magnetic resonance imaging (MRI) quantification, already used in other organ systems, to assess for infection/inflammation, in women with and without PPROM aiming to explore its use as a biomarker for inflammation within the feto‐placental compartment in vivo.


| INTRODUC TI ON
Preterm birth (PTB) is defined as delivery before 37 weeks of gestation and occurs in up to 8% of pregnancies 1 in the UK. It is associated with fetal morbidity and mortality both in the neonatal period and beyond, particularly at very preterm gestations and where placental inflammation (chorioamnionitis) is present 2 . As a result, PTB also constitutes an important burden for public health. 3 Approximately 40% of cases of PTB are associated with preterm prelabor rupture of the membranes (PPROM). 4 PPROM before 37 weeks of gestation complicates 3% of all pregnancies in the UK. 5 The median time between membrane rupture and delivery in these cases is 7 days but where delivery has not occurred expectant management is advocated in the absence of signs of maternal infection, to reduce the morbidity and mortality associated with preterm delivery before 37 weeks. [6][7][8] Infection is a significant complication of expectant management, and can often occur in the fetus in the absence of clinical signs in the mother. 9 Neonatal morbidity, including sepsis, cystic periventricular leukomalacia, intraventricular hemorrhage, and later development of cerebral palsy, are significantly higher among pregnancies with PPROM complicated by infection as assessed by chorioamnionitis in the placenta after delivery. 5,10,11 Chorioamnionitis is also known to alter the immune profile of the infant at birth. 12 Up to 71% of cases with PPROM have histological evidence of chorioamnionitis on placental analysis, particularly at early gestations, yet 30% can be subclinical, without signs of maternal fever, fetal tachycardia, uterine tenderness or purulent discharge. 9 Expectant management may therefore increase risks for both mother and baby if signs of infection are not overt.
Placental histopathological assessment only allows a retrospective diagnosis of chorioamnionitis and does not facilitate timely antenatal intervention such as administration of corticosteroids and/ or early delivery or in utero transfer. Advanced functional magnetic resonance imaging (MRI) techniques provide a means of quantifying tissue changes and have been used in the gastrointestinal tract 13,14 and heart 15 to discriminate between acute and chronic phases of inflammation in vivo. Placental MRI has been pioneered in normal pregnancies during hyperoxygenation and normoxygenation 16 in the assessment of conditions including fetal growth restriction 17 and preeclampsia 18 revealing distinct phenotypes. Techniques include magnetic resonance relaxometry, targeting the paramagnetic properties of deoxygenated hemoglobin to gain insights into the oxygen concentration, 18,19 diffusion MRI to identify microstructural changes to the villous trees and flow in the inter-villous spaces, 20 as well as combinations of approaches leading to more distinct descriptions of placental compartmental properties. 21,22 This study aimed to assess the feasibility of MRI as a non-invasive antenatal assessment tool for evaluating intrauterine infection based on placental assessment. This could prove invaluable information for both the timing of delivery, and antenatal interventions, with the ultimate aim to minimize the morbidity/mortality associated with infection/inflammation in the context of PTB.

| Cohorts and clinical data collection
Women with PPROM before 34 weeks of gestation were recruited prospectively from a tertiary referral hospital in South London (Table 1). with lower T2* (mean T2* at 30 weeks 50 ms compared with 58 ms in controls) and higher fractional anisotropy (mean at 30 weeks 0.55 compared with 0.45 in controls).
These differences did not reach significance and there was substantial heterogeneity both in T2* and Apparent Diffusivitiy across the cohort.
Conclusions: This first exploration of functional placental assessment in a cohort of women with PPROM demonstrates that functional placental MRI can reveal a range of placental changes associated with inflammatory processes. It is a promising tool to gain information and in the future to identify inflammation in vivo, and could therefore assist in improving optimal timing for interventions designed to prevent fetal injury.

K E Y W O R D S
chorioamnionitis, histopathology, inflammation, magnetic resonance imaging, placenta, preterm prelabor rupture of membranes, preterm birth

Key Message
This study proposes novel functional MRI methods as a biomarker to assess signs of inflammation within the human placenta revealing a distinct phenotype in women with preterm prelabor rupture of membranes and placental inflammation.
Preterm prelabor rupture of membranes was confirmed on clinical grounds using speculum examination and biochemically using the rupture of membrane (ROM) test (ROM Plus ® Rupture of Membranes Test; Clinical Innovations) when consented. Inclusion criteria included 16-34 weeks of gestation, singleton pregnancy, not in active labor, and ability to give informed consent. Exclusion criteria were: diagnosis of gestational diabetes, preeclampsia, chromosomal abnormalities, any other diagnosis associated with placental insufficiency, maternal body mass index greater than 35 kg/m 2 , multiple pregnancies, metallic implants, and claustrophobia. Following assessment of eligibility, written consent was obtained and an MRI scan was performed on a 3T MRI system (Phillips Best).
One participant, scanned because of suspected PPROM, did not have a confirmed ROM test, had normal amniotic fluid at subsequent scans throughout and did not continue to leak fluid vaginally, so the decision was made to exclude her from any subsequent analysis. The number of days from MRI scan to delivery (see Figure 1A for a timeline), pregnancy outcome including gestation, weight at birth, fetal sex, Apgar scores at 5 minutes, and postnatal and maternal complications were recorded. Placentae were analyzed histopathologically using a structured assessment 23 and assessed both macroscopically and microscopically. Evidence for chorioamnionitis, umbilical cord vasculitis, funisitis, chorionic plate fetal vessel vasculitis, presence of thrombi, villitis, infarction, fibrin deposition, and maturation of the villi was recorded wherever available (see Figure 1B for the locations).

| MRI assessment
Women were scanned in a clinical 3T scanner using a 32-channel cardiac coil in supine position with adequate cushioning and positioning.
Regular verbal communication was maintained throughout the scan.
The scan time was limited to 60 minutes in blocks of 30 minutes with a break offered in the middle. Continuous monitoring of oxygen saturations, heart rate and blood pressure (in 10-minute intervals) was undertaken. Optimization of all sequences was performed previously to keep the acoustic noise below 98 db(A). Following localizer and preparation scans, anatomical scans of the entire uterus were performed using two-dimensional TSE sequences in five orientations. A B0 map was acquired and manual shimming using an in-house tool was performed focusing on the placenta. 24 Then, a two-dimensional multi-slice multiecho gradient-echo echo planar imaging sequence with four echo times was acquired to allow T2* mapping with the following parameters: Note: Only the second scan from the participant scanned twice was included in the statistics presented here. The numbers indicate mean ± SD.
F I G U R E 1 Schemata of (A) the timelines involved and (B) the localization of relevant inflammatory processes in the placenta. GA, gestational age; MRI, magnetic resonance imaging; ROM, rupture of the membranes field-of-view = 360 mm × (320-400) mm × (60-120) mm, resolution 3-mm isotropic, repetition time = 2.6 s, echo times between 11.6 and 182 ms, 1-2 dynamics. A combined diffusion-weighted relaxometry scan with optimized b-values, b-vectors and four echo times as previously described 19,22 was obtained.

| Data processing
All MRI data sets were assessed for overt fetal pathology. Placental data from all three considered functional modalities was manually segmented by two experienced placenta observers (authors JH, AH) keeping a conservative margin to the chorionic and basal plate to avoid inclusion of any non-placental maternal tissue or amniotic fluid. These segmentations included the entire placental parenchyma for the multi-echo gradient echo scan and the diffusion MRI.
Mono-exponential decay models were fitted to the data acquired for T2* maps using an in-house python script to obtain proton density and T2* maps: 10 random initializations were performed and the voxel-wise median value was used. The diffusion data were processed in a similar way to a previous study 19 to obtain maps of the apparent diffusion coefficient (ADC) and the fractional anisotropy (FA).
Quantitative values were obtained as the mean over the entire placental parenchyma. Additional histogram-based evaluation was performed for the T2* values resulting in skewness and kurtosis values.

| Ethical approval
The data used for this study were acquired as part of three ethically approved studies reviewed by the relevant ethical committees:

| RE SULTS
In total, 12 women with PPROM were successfully scanned (only the second scan from the one woman who underwent two scans has been included in the analysis). Eighty-seven control pregnancies were included for comparison. For the PPROM cohort, mean gestational age (GA) at MRI was 26 +5 weeks (range 19-33 + weeks), mean GA at delivery 29 +2 weeks (range 20 +1 -33 +3 weeks) with median duration from MRI to delivery 23 +2 days (range 3-106 days). For the control cohort, median GA was 29 +3 weeks. The pregnancy and birth outcomes for the PPROM cohort are given in Table 2.
Complete anatomical and relaxometry data were obtained and processed in all cases. Diffusion data were obtained in all cases, but judged of sufficient quality in only 10 cases. Histopathology results are given in Table 3

| Visual appearances
Anatomical changes were apparent in the sagittal displays for PPROM cases P1 and P5 with large areas of hypointensity close to the chorionic plate (marked with green arrows), that were not seen in control placentae. Smaller hyperintense areas were observable (pink arrows) in most cases but these were not present in the control cases of similar gestation. 25 Control T2* maps 25 show an increasingly lobular appearance with more pronounced circular hyperintensive areas surrounded by hypointense structures over gestation. These changes typically occurred homogeneously over the entire placenta. However, the T2* maps of the PPROM cases reveal pronounced, focal large areas of low signal-as indicated with green arrows in PPROM placentae P1, P2, P5, P6, P9, P12, and P13-which were not as visible in the anatomical images from the majority of women. P10 and P11, from the same woman, do not display these signs.     There are several recent studies using MRI to study PTB, but these have focused on volumetric assessment of fetal organs. They found a reduction in lung volume 26 and changes in the size of the thymus gland-known to play an integral role in the development of the fetal immune system and suggested as a marker of the fetal inflammatory response. 27 Previous efforts to assess the placenta specifically detected changes of a bank-like T2-weighted hypointense signal and diffusion-weighted hyperintense signal changes associated with a diagnosis of chorioamnionitis in three of six patients. 28 This current study goes beyond anatomical assessment and includes functional MRI properties of the placenta, which could complement fetal findings.

F I G U R E 2
Imaging results of all preterm prelabor rupture of membranes (PPROM) cases P1-P13. P10 was omitted and the second scan of this woman, P11, was included. For each case, the anatomical images are displayed in the first row, coronal placental view, sagittal placental view and sagittal cervix view. The second row gives the functional T2* data in coronal and sagittal view matched to the anatomical data. Pink arrows identify regions of hyperintense small dots. Green arrows identify areas of reduced T2*. In the right column three exemplary controls are shown at similar gestational age The use of T2 and T2* relaxometry as a means of discriminating between acute and chronic phases of inflammation has previously been proposed in other organs such as the heart and the kidney 13   However, the cited changes in structure, the changed water content as well as any other changed tissue property might influence the transverse relaxation rates. cervical changes among other, and severity of chorioamnionitis, which initially affects the maternal compartment before progressing to the fetal tissues, resulting in differing characteristics on placental histology. [29][30][31] The maternal inflammatory response is characterized first by microbial invasion of the chorion, amnion, and chorionic plate before progression to epithelial necrosis of the amnion. 23,29 Where this process is localized to a specific area of the placenta (such as for case P2) mean T2* is not as low as in cases where the invasion is more diffuse (cases P1  33 and so facilitate this in the future. Efforts to display the placental tissues in an agreed coordinated system could further add to the specificity of the localization of lesions. 34,35 To usefully translate these results, following a larger observation study, would be a randomized control trial to assess whether altering the timing of delivery in PPROM in response to MRI biomarkers of infection/inflammation could result in a reduction in neonatal morbidity and mortality. Combined analysis with biomarker-based serum scores might further enable the development of individual risk scores. 36,37 Phenotyping different presentations in more detail together with histopathological assessment will allow additional insights into the cascade of events starting from the ascending infection through the layers of the placenta to widespread fetal infection.

| CON CLUS ION
Functional placental MRI reveals a range of placental changes, associated with inflammatory processes confirmed on subsequent histology. It shows promise as a tool to noninvasively identify inflammation in vivo, and could therefore assist in improving optimal timing for interventions designed to prevent fetal injury, such as antenatal corticosteroids and magnesium sulfate and the need for delivery and/or in utero transfers where indicated.

CO N FLI C T O F I NTE R E S T
AHS is the chief investigator on a number of trials funded by NIHR and charity sources related to preterm birth prediction and pre-