Late unexplained stillbirth is associated with maternal going to sleep position. MRI studies have elucidated the physiological effects of the maternal supine position on uteroplacental blood flow in normal late gestation pregnancy. This presentation will review the results of position studies as a model for investigating vulnerable pregnancies where the fetus may not be able to adapt to hypoxic stimuli. The direction that future studies may take will also be suggested.
Peter Stone is Professor of Maternal Fetal Medicine in the University of Auckland and formerly Head of Department in Obstetrics and Gynaecology. His postgraduate training was in Britain, gaining a Doctor of Medicine based on Doppler studies in fetal growth restriction from the University of Bristol. After working in Wellington at the University of Otago for 11 years, where he set up the maternal fetal medicine service, he moved to Auckland in 1998. and was instrumental in the development of the New Zealand Maternal Fetal Medicine Network.
He has been a member of a number of Ministerial advisory groups including that for Down Syndrome screening and is a member of the Technical Advisory Group for Antenatal and Newborn Screening in the National Screening. For 6 years he had a UNFPA consultancy to assist in the development of maternal fetal medicine in Mongolia.
He is currently a member of the Neonatal encephalopathy working group specifically exploring the introduction of cord lactate testing at birth to reduce adverse neonatal outcomes.
Research interests include fetal welfare assessment and stillbirth, obstetric ultrasound, early pregnancy development and trophoblast biology and the detection and assessment of fetal growth abnormalities.
Current research projects include assessing maternal positional effects on fetoplacental oxygenation and investigation of circadian patterns of fetal heart rate in normal and growth restricted pregnancies
Professor Stone has disclosed that she does not have any real or perceived conflicts of interest in making this presentation.
Amy Wetjen: Dr. Peter Stone is Professor of Maternal Fetal Medicine in the University of Auckland and formerly Head of Department in Obstetrics and Gynecology. His postgraduate training was in Britain, gaining a Doctor of Medicine on Doppler studies in fetal growth restriction from the University of Bristol. After working in Wellington at the University of Otago for 11 years, where he set up the maternal fetal medicine service, he moved to Auckland in 1998 and was instrumental in the development of the New Zealand Maternal Fetal Medicine Network.
He has been a member of a number of Ministerial advisory groups including that for Down Syndrome screening and is a member of the Technical Advisory Group for Antenatal and Newborn Screening in the National Screening. For 6 years he had a UNFPA consultancy to assist in the development of maternal fetal medicine in Mongolia. He is currently a member of the Neonatal Encephalopathy Working Group specifically exploring the introduction of cord lactate testing at birth to reduce adverse neonatal outcomes.
Professor Stone’s research interests include fetal welfare assessment and stillbirth, obstetric ultrasound, early pregnancy development and trophoblast biology, and the detection and assessment of fetal growth abnormalities.
His current research projects include assessing maternal positional effects on fetoplacental oxygenation and investigation of circadian patterns of fetal heart rate in normal and growth-restricted pregnancies. Professor Stone’s presentation is titled, Exploring the Pathophysiology of Late Stillbirth: How MRI Studies of the Maternal and Fetoplacental Circulation May Help.
Professor Peter Stone: Welcome to this webinar as part of the Star Legacy’s meeting this year. It’s a pleasure to be able to share some of the findings about MRI studies with you. The objective of the presentation is really to discuss our findings, which relate to the fact that late unexplained stillbirth is associated with maternal going to sleep position. MRI studies have elucidated the physiological effects of the maternal supine position on uteroplacental blood flow in normal late gestation pregnancy.
This presentation will review the results of position studies as a model for investigating vulnerable pregnancies where the fetus may not be able to adapt to hypoxic stimuli. The direction that future studies may take will also be suggested, and during the presentation, some possible future research directions are suggested
This slide reminds us of the size of the problem in the so-called high-income countries. The distinctions about income are really related to presumed etiology of stillbirth amongst other reasons, though it should remind us all that one of the common causes of stillbirths in some countries such as Mongolia, where I’ve worked, is syphilis. This unfortunately has led to an increasing number of stillbirths in other countries including New Zealand. The point is that the evidence based on clinical and pathological classifications of etiologies would suggest that many, if not the majority of late stillbirths are preventable.
Our studies and what I will present in this talk are the results of at least a 10 to 15-year research journey. By way of background to our MRI studies, it was a way back in 2006 to 2009 that members of our group undertook a clinical epidemiological study. A midwife interviewed women who had recently had a stillbirth and these were matched against women who had had a healthy outcome.
Professor Ed Mitchell, that many of you will know, had done work on safe infant sleeping. He had suggested to Tomasina Stacey, who was our research midwife, to ask out sleep-disordered breathing and maternal sleep positions in pregnancy. That came up in the subsequent analysis and as an independent risk factor associated with the risk or likelihood of late stillbirth.
To help verify her findings, we repeated the study that Tomasina had done as a multi-center study in New Zealand and then others around the world also had similar findings. On this slide, you’ll see that we had initially showed a twofold increased risk of stillbirth with maternal supine going to sleep. We confirmed this as a threefold risk in our multi-center study. An Australian study showed higher risk, and these findings were confirmed later in the UK and in India, and earlier there’d been a slightly different study in Ghana.
Then another one of our research midwives, Robin Cronin, went on and performed an individual patient data meta-analysis from data available from 826 cases where there’d been a stillbirth, and over 2,200 controls. She undertook this large project as part of her PhD. These are her findings, but the important one really relevant to tonight’s talk is where the red arrow is. The findings that have been recently published confirm the independent association between going to sleep position and the risk of stillbirth.
It would appear that compared with going to sleep on the left side, which is the referent in her work, supine sleep position is associated with a nearly threefold risk of stillbirth. There have been little difference between the left and the right side. That was a variable that we were interested in because we did find a difference in our first study that Tomasina Stacey had done. We also thought that given the asymmetry of the human vascular anatomy, that we might find a difference in both the epidemiological and MRI studies.
We had, in fact, performed a small MRI study prior to the IPD here, comparing left versus right in late pregnancy. Whilst finding some significant differences in vena caval blood flow at the level of the renal veins, otherwise, we found no differences in maternal cardiac output or other parameters when comparing left and right, when previously we had shown differences between left and supine.
Now, many of you will be aware that there are a large number of associations or risk factors for late stillbirth, but the question remains, what are the mechanisms that may explain at least some of these epidemiological factors? Also, some of these may operate through a variety of associations and may not be independent as maternal position is.
Some years ago, Jane Warland and Ed Mitchell proposed the triple risk model, which sought to group or categorize the origin of factors that may coalesce and result in fetal demise. We sought to investigate why the maternal going to sleep position was associated with an independently increased risk of stillbirth. Not only in our first single-center, then the multicenter study, but also in subsequent IPD, all of these studies had confirmed the going to sleep position as an independent risk factor, but there had been no explanation as to possible mechanisms for these findings. Some of these risk factors as described by the previous slide and in the triple risk model may operate through a variety of associations and they may not be independent.
We set about to investigate maternal position effects. First, we set out to examine the effect of maternal position on maternal hemodynamics, seeking a reason for why going to sleep supine might be associated with what we presumed to be hypoxic or an acidemic model of fetal death. Later on in this presentation, I’ll show you results from our work on uteroplacental blood flow and fetoplacental oxygenation, all work involving various modalities of MRI.
In the first studies, we recruited healthy pregnant women between 34- and 38-weeks gestation with a singleton pregnancy. The starting position in the MRI scanning either left lateral decubitus or supine was randomized, and we imaged at the aortic root or the level of superior vena cava, and the level of the renal veins, and at the aortic bifurcation.
Following this, once we had completed studies in over 10 women, but we had complete results in 10, we then repeated the process in a different set of women, this time comparing left lateral with right lateral positions. In all our flow velocity and later MRI studies that we have conducted, we performed intra and inter observer studies after training of the investigators.
We randomized the maternal starting position, as I’ve mentioned. Throughout the MRI scanning, we monitored the maternal blood pressure and heart rate using MRI safe devices. We undertook phase contrast imaging, and then we used the Siemens proprietary package called Syngovia to calculate blood flow velocities.
The illustrations here show the effect of position on the inferior vena cava, and here a woman lying in a left lateral position, you can see that the inferior vena cava outline in blue is large. Whereas when the woman is lying supine, the inferior vena cava is remarkedly compressed. These images are from the same woman just turned into different positions.
We went on to show that the cardiac output was reduced when the woman was supine. What we did show was that the cardiac output was maintained at least in part by the collateral venous system draining by the azygos vein into the superior vena cava. A research question that remains is whether venous abnormalities may affect the function of the collateral veins, especially the pelvic and paravertebral veins. If so, would these venous anomalies place a woman at increased risk of reduced cardiac output in late pregnancy when supine and therefore potentially at risk of stillbirth?
Our key results here, they showed that the supine position resulted in a 16% production in maternal cardiac output. Also importantly, at the abdominal aorta bifurcation into the common iliac, there was a 32% reduction in blood flow when the woman was supine.
The question was, well, how was cardiac output maintained with vena cava compression? Here we showed that when the woman was supine, the flow in the azygos vein, which drains into the superior vena cava and then to the right atrium was increased to 220%. This was the first time that this had been shown and MRI’s really the only non-invasive tool which would allow us to derive these results.
The research questions then are, well, not only are there women potentially with venous abnormalities that this adaptation, when a woman is supine, could not occur, but also at what gestation does the maternal position cause these hemodynamic changes because our studies were done in late gestation. Our results to some extent have been validated by a more recent study from a group in London who have shown changes in the spinal venous plexus in the pelvis and draining into the paravertebral veins when the woman lies supine.
The questions that we then asked, well, we’ve shown these changes in maternal hemodynamics, but are these changes important enough? Do they have important impacts on the uteroplacental circulation and oxygen transfer to the fetus? The MRI as a functional tool, and also being non-invasive, may well be very appropriate technology to attempt to answer these questions.
After presenting the hemodynamic results internationally, we began a collaboration between Auckland and Anna David and Andrew Melbourne’s group at University College Hospital and King’s College in the UK.
The UK group had described a multi-parametric MRI model, which was sensitive to fetal blood oxygenation, and they coined the acronym DECIDE which is a mouthful, but it’s Diffusion-rElaxation Combined Imaging for Detailed Placental Evaluation. Really in brief, what that does is that DECIDE allows for the simultaneous assessment of oxygen saturation and perfusion or diffusion in both maternal and fetal perused tissues. It seems to be the only technique at present that can do this.
Now, one of the reasons that we were very interested in being able to use this technique was that in a series of overnight sleep studies that we had done some time ago, we had shown that the supine position was associated with the change in fetal behavioral state, such that when a woman moved from lateral to supine, the fetus rapidly changed from an active to acquiescent state. We hypothesized this was due to a mild hypoxic stimulus and was an adaptive response to conserve oxygen. Hence, we were very interested to see whether using the DECIDE technique we could actually quantify the effect of assuming a supine position on oxygen transfer and oxygen saturation in the fetus.
Now, for those of you that are interested in technical data, these are the two publications that describe the technical details behind the DECIDE model paper by Andrew Melbourne in 2018 and by Rosalind Aughwane in 2019.
We undertook a study in late gestation healthy pregnancy. We chose this gestation period to correspond to the period in which we’d done our hemodynamic studies. Again, we took singleton healthy, normal pregnancies with a pre-pregnancy BMI of less than 30 with no other maternal comorbidities. The imaging in each position of supine and left lateral took approximately 25 minutes. We used the 1.5 Tesla MRI scanner. We randomized the maternal positions again. We used the DECIDE process to analyze the imaging of the placenta. We used phase contrast imaging of the internal iliac arteries to look at the flow into the placenta, we looked at phase contrast imaging of the umbilical vein to look at the flow out from the placenta into the fetus, and we measured maternal heart rate and blood pressure throughout these studies.
The blood flow was calculated using Syngo Via Software. This is a proprietary software. It was actually validated in Auckland some years ago by Professor Young and his colleagues. We used DECIDE to look at oxygenation. In our analysis, the two observers who independently analyzed all the data were blinded to maternal position.
The DECIDE parameters produced metrics for perfusion and oxygenation. Diffusion weighted imaging relates to perfusion, and T2* relates to oxygenation. We examined both the uterus on the maternal side of the circulation and the placenta, both maternal and fetal circulations.
The diffusion analysis produces these masks for the DECIDE parameters. This relies on what’s called segmentation of the placenta which is defined by the investigator by drawing a line around the object in question. In this case, the placenta. All our segmentations are done in duplicate by two observers to ensure good inter observer correlations.
In this slide, I show you a summary of the findings investigating maternal blood flow to the uterus, the umbilical venous flow and oxygen delivery, which we have termed flux, and I’ve colored in the boxes with relevant results. We found that there was no dominant feeder artery or internal iliac artery. Both of them showed high flow in the left lateral with marked reduction when the woman became supine, overall nearly a 24% reduction in maternal blood flow into the uterus. Then placental flux, which is a measure of oxygen movement within the placenta showed a significant 6.2% decrease when the woman was supine. Then the delivery flux, which is an indicator of oxygen transfer to the fetus, was reduced by 11%.
Now, I’d like to point out that the group in Toronto who are also doing novel work in this area reported results from the uterine arteries. Certainly, in our hands, we believe that these are below the usual resolution of the MRI, and as such, can’t be reliably imaged. We found that the internal iliac arteries were the main feeders to the uterus, and that we could visualize these in all cases.
In summary, these first studies showed that supine position in late pregnancy is associated with significant reductions in internal iliac arterial blood flow and placental flux, that’s oxygen movement across the placenta, with a reduction in fetal oxygen saturation. We would interpret this as showing that the reduction in oxygen saturation is likely to be a stressful vulnerable fetus, but other mechanisms may also be involved, and in the remainder of the talk I will discuss some of these.
One of the important associated factors which need elucidation as I’m sure many of you will be aware is fetal growth restriction. Robin Cronin had shown in the IPD that there are other factors which could be associated with fetal vulnerability, and those are listed here. These are likely to be independent and additive and that would of course fit with the triple risk model that I’ve already shown you.
These data from a paper by Jason Gardosi published in 2013 show the impact of fetal growth restriction on stillbirth rates. Fetal growth restriction is arguably the single biggest cause or association with stillbirth in otherwise normally form fetuses. In this slide, looking at stillbirth in general, and then with no fetal growth restriction, fetal growth restriction and up to a 15-fold risk of stillbirth reduced to where the fetal growth restriction is detected, but where it’s not detected, the risk of stillbirth is even higher.
Having shown an early MRI results in healthy, normal pregnancy that position had an effect on a risk of stillbirth, we were interested in testing our model of a fetal stress, so where there is position change in growth-restricted pregnancy. These are Gardosi’s results just shown graphically, illustrating not only the importance of the detection of fetal growth restriction, but the other point he makes is the effect of gestation on the impact of fetal growth restriction on stillbirth.
In the IPD, Robin Cronin showed that for all fetal growth restriction, maternal position was again an important independent risk factor, but this time, even more important than in normally growing fetuses– here we have the non-SGA. Left being the referent position with the 2.5-fold risk of stillbirth should the woman go to sleep supine. That increases to 15 when the fetus is small for gestational age.
It’s important to note that to date the epidemiological studies have assessed risk based largely on birth weight. That is, SGA, small for gestational age, which is not the same as fetal growth restriction. Though, of course, there is an overlap, but the SGA category will miss growth-restricted fetuses who are above a centile cutoff, which is typically taken around the 10th centile.
Using a secondary analysis from the IPD, Ngaire Anderson in our team showed that there was a 10% reduction in mean birth weight in those pregnancies where the mother recorded a supine going to sleep position. This was confirmed both on the use of the INTERGROWTH charts as well as the customized birth weight centiles. It would seem that maternal going to sleep position has not only a direct effect on stillbirth risk, but it also one of the important risks associated with being growth-restricted.
We can consider maternal supine position to be an acute hypoxic stress. So even a normal pregnancy. But we know that the fetus has a number of adaptations to hypoxia. If we look at the fetal circulation in the slide, we note that there are a number of special shunts that move blood away from the fetal pulmonary circulation into the systemic circulation.
In hypoxia, both in animals and in humans, it’s been well shown that a number of these shunts change in hypoxia. In particular, the ductus venosus will change the amount of blood shunted away from the liver. In hypoxia, up to 25% of the umbilical venous blood, which would normally be passing through the liver, will be shunted directly to the heart.
The result of this, of course, is that more oxygenated blood then crosses through the foramen ovale to the left side of the heart, and then to the brain. At the same time we see dilatation of the middle cerebral artery, which can be shown on Doppler. The summary result is that in hypoxia, the fetus can make protective adaptations to improve blood supply to the vital organs of the brain, the heart, the fetal adrenal, and of course, back to the placenta.
Knowing that these effects happen in hypoxia and knowing that we’d shown that the maternal supine position reduces placental oxygenation in normal pregnancy, we added liver to our DECIDE outputs. One of our team, Rebecca Gandhi, segmented the fetal liver in our position studies of healthy, normal pregnancy and assessed the effect of maternal position change on blood flow and oxygenation in the liver.
We were particularly interested to investigate whether the hypoxic effects of maternal supine position were sufficient to cause fetal centralization or brain sparing in healthy normal pregnancy. We wanted to investigate the effect of maternal position on myometrial and fetal hepatic blood flow and oxygenation.
We used a program called ITK-SNAP to perform segmentations of the liver and each patient had two DECIDE images, one supine, and one on the left lateral. The segmented sections of the liver then underwent the DECIDE analysis. The women were of course their own controls, and so we were able to perform PT tests to compare the results from left lateral and supine. These are the DECIDE parameters, again, that we looked at as we had done before. This shows in green, the placenta, in blue, the fetal liver, and the red is the myometrium. The maternal uterine myometrium.
Our results, in summary, showed that maternal supine position was not associated with any change in maternal perfusion of the myometrium overall though there was a decrease in myometrial oxygenation but there was no change in fetal liver perfusion and no change in fetal liver oxygenation in these healthy pregnancies. This really supports the concept raised many years ago, I believe, by Bryan Richardson in Canada about the fetal oxygen margin of safety. In other words, the fetus has a tolerance to mild hypoxia when the pregnancy is normal.
The research questions that really crop up from this would include the effect of position change as a hypoxic stress in the growth-restricted fetus, and again, the effect of gestation on this.
A summary now, adding in our results of the liver, is that we can show that although position change is associated with reduction in placental flux, umbilical blood flow, oxygen delivery flux to the fetus, and although there are small changes in liver fluxes oxygen delivery to the fetal liver, those changes were not significant and clearly tolerated in these healthy pregnancies.
Now, the group in London, Rosalind Aughwane and team have reported on a cohort of fetuses with severe early fetal growth restriction. These were scanned only in the left lateral position, but the fetal growth-restricted group had a mean gestation of 27 weeks, controls 29 weeks. Looking at these compartments, so the chorionic vessels, the villous vascular tree, the intervillous space, and maternal vessels, they showed lower fetal placental oxygen saturation in the growth-restricted fetuses, actually a saturation of only 56% compared to 75% in healthy controls.
Of interest though in the data showing here, they broke their subjects down into groups according to the Doppler ultrasound that they had done at the same time. They found that fetal perfusion, maternal placental perfusion, and fetal oxygen saturation was lower, not only in the growth-restricted pregnancies compared normal, but it showed a relationship with the uteroplacental and fetal Doppler recordings.
What that means is that the measures obtained in the DECIDE parameters appear to correlate with disease severity as indicated by Doppler. Now, DECIDE is an early model, which aims to describe the path of physiology and non-invasively measures of placental function and fetal oxygenation. What it may do is help define the trajectory of the disease and inform optimal timing of delivery. Here, we’ve got images from DECIDE. Here’s a normal placenta, which appears remarkably homogeneous. Here are two placentae in very growth-restricted fetuses. This is a very, very small placenta, and this one is markedly heterogeneous in its character compared with the normal. This suggests placental disease.
The London group concluded that MRI assessment of the placenta may be able to elucidate placental disease, and in doing so offer a better prediction of when the placenta may fail. Clearly, more work is needed, and our physiological model using maternal position change may actually offer a way of testing just how severe the placental disease is in fetal growth restriction. Obviously, though, the big challenge is that the growth-restricted fetus has to be detected in the first place as Gardosi’s data showed, but that is really a different topic for another day.
I will leave you with some possible research questions which have derived from this work to date. What is the effect of maternal position change in fetal growth-restricted pregnancy, and the effect of gestation? Can MRI help solve the issue of the heterogeneity of fetal growth restriction? Can MRI really define the trajectory- [sound cut]
This presentation was part of the Stillbirth Summit 2021. This individual lecture will be awarded .75 hours of continuing education credit to include viewing the lecture and completing evaluation and post-test. Once received a certificate will be emailed to the address you provide in the post-test. If you did not register for the Summit WITH Continuing Education, you can purchase the continuing education by clicking here. This purchase will provide you access to all Stillbirth Summit 2021 lectures including continuing education credit. There is no charge for viewing the presentation.
Kristin Naylor: Hello, I’m Kristin Naylor. My daughter Abby died July 13th, 2018 from a hypercoiled umbilical cord accident. We miss her and we love her so much. In her memory, I am honored to introduce Professor Stephen Tong. He’s an academic OB/GYN based in Melbourne, Australia. He holds a Bachelor of Medicine, a Bachelor of Surgery, and PhD from Monash University, and he is Fellow of the Royal Australian and New Zealand College of Obstetricians and Gynecologists.
Over the past decade, he has led the translation of new treatment for ectopic pregnancy and preeclampsia, from laboratory concept to phase I through III clinical trials. He also leads a program of research that is hunting for a new blood test that may help women avoid a stillbirth. Professor Tong’s presentation is titled, The Hunt for a Blood Test to Reduce the Risk of Stillbirth.
Professor Stephen Tong: Greetings. Thank you very much for the kind invitation to present at the Stillbirth Summit. I’d like to thank Lindsey and the Star Legacy Foundation for this kind invitation. It looks like the Star Legacy is indeed an impressive organization. It incorporates looking after women and their families touched by stillbirth, you provide education to clinicians, and you support research into stillbirth.
It’s my pleasure to present to our work, which is to hunt for a blood test to reduce the risk of stillbirth. I would have been very keen to visit the US and to get to know the impressive team at the Star Legacy Foundation, but perhaps in another time.
I’m a named inventor for a Patent for SPINT1, which is a marker I’m going to tell you a bit about today, and I hold a licensing agreement with an industry partner
For the first eight or nine months of its existence, the unborn baby is critically dependent on one organ to survive and to thrive, and that’s the placenta. If the placenta is not doing a good job of delivering nutrients and oxygen to the unborn baby, the baby gets stressed. It therefore mobilizes and channels all its existing fuels from growth and development to just survival, and consequently the growth trajectory of the baby falls and we have a situation of fetal growth restriction.
Unfortunately, a small number of babies that are growth restricted continue to do more poorly where the placenta works less and less well. In a small number of babies that are growth restricted, unfortunately, a stillbirth is the result. Fetal growth restriction has association with an increased stillbirth risk.
I’d like to present to you some representative epidemiological data just to illustrate this point. This is data of around 662,000 pregnancies in my state in Australia, the state of Victoria. You can see this is the risk of stillbirth plotted against gestation in weeks. You can see in the gray line that the risks of stillbirth for those over the 10th centile is pretty low throughout your pregnancy. Then it rises and lifts quite sharply from 38 weeks to 40 weeks of gestation and increases particularly sharply over 40 weeks gestation.
In contrast, those that are small for gestational age or growth restricted that is defined under the 10th centile birth weight, the overall risk of stillbirth constantly is three to fourfold higher right across the pregnancy. This risk lifts sharply to an alarming degree once the babies hit term gestation. I’d like to just note two takeaways from here. If you are small in utero, you have a three to fourfold increased risk of stillbirth, and also that the risk of stillbirth for all pregnancies lifts sharply once babies hit term, and certainly post-term gestations,
This is a bit more epidemiological data. It comes from the UK. It’s from Jason Gardosi’s group. This is the risk of stillbirth in 92,000 pregnancies. The overall risk in this UK cohort of stillbirth is 4 in a thousand. It was only 2 in a thousand if the babies weren’t growth restricted, again, defined as under the 10th centile. If they were under the 10th centile, the risk of stillbirth lifts and quadruples to 16 in a thousand.
What is informative about this data is the following. The team then split this cohort that were growth restricted into two groups. If you were growth restricted, and it was recognized during the pregnancy, that is, fetal growth restriction detected, the stillbirth risk was 9.7 in a thousand. If there was the presence of fetal growth restriction, but it was not recognized during the pregnancy, as in, the baby was small, but it was not appreciated that it was small, the stillbirth risk then doubles to 19.8 in a thousand pregnancies.
This means that if the baby’s small, but it’s recognizing in time while women are still pregnant, we can put in measures to protect the baby. We can watch the baby very carefully with ultrasound monitoring, for instance, and we can time birth to prevent a stillbirth because if those babies are vulnerable, if you deliver the baby, then the stillbirth risk evaporates if the stress is caused by the placenta.
In fact, this team concludes that the single largest risk factor for stillbirth is unrecognized fetal growth restriction, and that preventative strategies need to focus on improving antenatal detection. You think, where’s the problem? Can’t we detect all babies that are small, we can therefore deliver them and watch them very carefully?
We are far worse at picking babies that are small in utero than a lot of people would think. It’s a time-honored approach to try and detect small babies by putting a tape measure on the tummy during antenatal or prenatal visits. The clinician sees the patient, we put a tape measure on the abdomen and we try and guess relative gestation whether the baby’s tracking small or not. Then the ones that we think are small, we send them for an ultrasound or selective ultrasound.
Unfortunately, the tape measure and selective ultrasound measure of picking small babies only detects about 20 to 30% of small for gestational age babies. That is babies under the 10th centile. It performs way worse than a lot of us think. I suspect it performs way worse than a lot of OB/GYN or midwives think. It’s a concern. You think, “Okay, if that doesn’t work well, then perhaps we can just perform ultrasounds on everyone at say 36 weeks gestation. Surely the ultrasound can bullseye a small baby if the baby is indeed small.”
One of the most important studies that examined this looking at 4,000 ultrasounds under 36 weeks was the POP study published in Lancet by my friend, Gordon Smith in Cambridge. It found that if you ultrasound everyone at 36 weeks, it only detected 57% of babies that are small, missing over 40% of babies at a small. These were the best ultrasound sinologists in research settings. In the real world setting where there may be perhaps variable skill and variable quality, the real-world detection more sits at 50% and there’s some reports with Harvard just under 50%. Ultrasound just picks just over half of babies that are small. We do have a problem. There is a large chunk of babies that are small, they go undetected, they have an increased stillbirth risk, and sadly, a few of them will succumb.
We are on the hunt to identify a blood test. Our vision is to develop a blood test at 36 weeks gestation. We’ll be able to pick babies that are small, but also hopefully assign a low risk of babies that aren’t small, and we can leave them alone and decrease medical interventions. But the ones that are small, we can monitor them and time birth and deliver them before stillbirth happens. That is our very focused goal, and that’s what we have a large research program for indeed currently funded with a large block funding from our Australian government to try to make this possible.
We have a goal. We want to develop a blood test at 36 weeks to pick small babies, but our next problem is what befalls most people trying to develop new diagnostics. You’ve got millions and millions of proteins and all sorts of molecules swirling in the bloodstream. How do we even choose? How do we start to identify a molecule which may be able to be used as a biomarker to pick a small baby? We’ve really landed on a very simple strategy to do this and one which has been yielding fruit for us.
Let’s just think of other branches of medicine. If we want to pick a problem with the liver, we measure liver enzymes, which are proteins, which are highly expressed in the liver and hardly expressed in other tissues of the body. If we want to pick whether there is a prostate problem, while it’s not the cleanest screening tests, we would measure the prostate specific antigen. This is a protein which is highly expressed in the prostate and little expressed in any other tissues in the body. If we want to pick whether someone’s had a heart attack with chest pain, we’ll measure cardiac enzymes such as troponins. These are proteins which are highly expressed in the heart and little expressed elsewhere.
If other branches of medicine do this, then we reasoned that very simply that to pick problems with the placenta, we should focus on screening proteins which are highly expressed in the placenta and express little elsewhere. As it turns out from our bioinformatic hunt, we found a treasure trove of 200 to 300 proteins which are highly expressed in the placenta and either little expressed elsewhere, but certainly the highest expressed in the placenta compared to any other tissues in the body.
We therefore reason that if there are situations where the placenta is not working well, the levels of these proteins may change in the bloodstream. We may not understand the biology why they change, but that wouldn’t matter. As long as they are consistently changed, then we could use this as a biomarker to pick that the baby is small and a possible increased risk of stillbirth.
I’d like to just show you, out of interest, how we pick these molecules and just to highlight an example of what we’re talking about. I’d like to show you data from this online resource called BioGPS. This is publicly accessible. Anyone can search this. You can search your favorite protein and you choose protein X, and it shows you all the different tissues in the human body and the level that your favorite gene or protein is expressed.
I chose randomly a protein which was completely in obscurity and hardly anyone knew about it three years ago, but in the advent of the pandemic sadly everyone’s heard about it. This is the angiotensin II receptor. This is the receptor which COVID-19 latches upon in the lungs. You can see that while it’s expressed in the lung, this angiotensin II receptor is expressed in many other tissues in the body. You may not be able to read this, but you’ve got pancreas, you’ve got prostate, lung, placenta, colon, liver, heart. It is ubiquitously expressed in many tissues in the body.
In contrast, let me just show you the same BioGPS readout for another search, which is HCG, human chorionic gonadotropin. This is the pregnancy hormone. It’s what is measured in the strips which women wee on in hope that they’re pregnant. It’s the molecule that we measure in the bloodstream to confirm women are pregnant. It’s a molecule specific to the placenta. Here’s the BioGPS readout. Of all the human tissues in the body, it’s only expressed in the placenta.
This has been our strategy. We look for many which are similar to this, highly expressed in the placenta and little expressed elsewhere, and we found a hit list of 250 to 300 potential biomarkers that we work through one by one to see whether any are able to flag a baby that’s small. We have a biomarker strategy, but to hunt and to test whether they may be able to predict or identify babies that are small, you need large human cohort samples to be able to test the hypothesis.
We undertook a large collection, which was a prospective cohort of 2000 participants which we call the FLAG study. We collected 2000 blood samples. We collected samples at 28 weeks, but specifically 36 weeks, which is where we would aim to develop our test. We split our FLAG cohort, our 2000 participants, roughly in half. A discovery set and a validation cohort.
In the first thousand samples, the discovery set, we then screened these 250 molecules one by one, and we split our cohort further within the discovery cohort into whether they were small for gestational age. As in, they ended up delivering a baby under the 10th centile, and the rest were controls, which were babies which ended up being over the 10th centile. We screened a lot of proteins, and I’d like to share with you our most exciting discovery. This is of a relatively obscure protein called SPINT1.
SPINT1 is an obscure molecule. It’s really not really been well studied at all. It sits on the surface of cells. It’s known to be a protease, so it efficiently eats up other proteins, specific enzymes such as hepatic growth factor. Very little is known about it. What is intriguing about SPINT1, what is already known is that the SPINT1 genetic knockout is embryonically lethal. They don’t survive past in utero stage. And there’s very severe placental abnormalities in the genetic knockout suggesting that SPINT1 may indeed play a very important role in placental biology.
Here is our graph of our initial discovery studies of SPINT1. You can see the small for gestational age or SGA under the 10th centile babies, the babies which ended up being birth small and had three or fourfold increased risk of stillbirth. Although in this cohort, they all came out alive. SPINT1 was significantly lower than the control group. That is pregnancies where we took bloods at 36 weeks, but they ended up delivering of a normal size. The difference here was very significant. The p-value was a very tiny 10 to the power minus 13. Very strong correlation with low SPINT levels at 36 weeks gestation in babies destined to be born small and presumably already small in utero when the bloods were taken.
Now, very importantly, in biomarker studies, you need to validate your results. Otherwise you overcall findings. We then ran SPINT1 in our validation samples, and we’re happy to note that SPINT1 levels were confirmed to be very significantly far lower than the control group.
Placental growth factor is currently before SPINT the circulating molecule with the strongest association with the placental insufficiency. Therefore, we measured placental growth factor in our validation group, and we found that placental growth factor while it was confirmed to be significantly lower in the small for gestational age cohort, the p-value is far lower and the area under the curve graph would suggest that it is far weak as a biomarker as SPINT1. This excited us because it suggested that SPINT1 may be a circulating molecule with the strongest association with placental insufficiency or small baby yet found.
This is a very famous epidemiological graph by Manning published in the ’90s. This shows you that with every step wise decrease in birth weight centile under the 10th centile, so 10, 9, 8, 7, 6 down to 1 and minus 1, you get an exponential increase in mortality and morbidity for the baby or the fetus. Therefore, if you are under the third centile, your risk of death in utero or succumbing in utero or serious morbidity even if you were born, was far higher than if you’re just tickling under the 10th centile.
If SPINT1 did have a strong link with placental insufficiency, we would expect that even among the cohort under the 10th centile, there should be a stepwise decrease in SPINT1 levels as you go down the birth weight centiles. This is indeed what we found where SPINT levels were far lower under the 3rd centile than they were between the 5th and 10th centile. This provides further scientific evidence, which links SPINT1 with true placental insufficiency.
We then tried to look for more scientific evidence to link SPINT1 with true placental insufficiency. We turned to preterm fetal growth restriction because this is perhaps the most severe variant of fetal growth restriction there is where the risk of stillbirth is extremely high. These very uncommon, but very serious situations, the maternal fetal medicine specialists need to watch these babies very carefully. We often have to deliver them preterm because the environment is too hostile, and they are very high risk of peril.
We indeed found that in placental samples from cases of preterm fetal growth restriction, the expression of SPINT1 was significantly lower. Then when we measured SPINT1 levels in the bloodstream of women with a preterm growth restricted baby on board, it was significantly lower than controls, and controls were blood samples taken at the similar gestations, but they were healthy pregnancies that were ongoing. SPINT1 levels were significantly lower in one of the most severe variants of fetal growth restriction, preterm fetal growth restriction.
We then went looking for even more evidence linking SPINT1 with true placental insufficiency, because if you develop the case more, it would suggest that SPINT1 may play an important role in placental insufficiency, and it makes the possibility that you can validate as a biomarker higher, but also the tantalizing opportunity that it may be a therapeutic target to save some of these babies. We’re not close to developing therapeutics to rescue growth restricted babies at the moment.
What we did is we turned to the laboratory. We exposed placental cells to hypoxia because we know that low oxygen is one of the parts to placental insufficiency and fetal growth restriction. Then we measured SPINT1 levels in the cells which were rendered hypoxic. We found that if you make cells hypoxic, put them in low oxygen in the petri dish, then SPINT1 levels downregulate and they go down.
We then looked at an animal model and we turned to our friend in Cambridge Amanda Sferruzzi-Perri. She has an animal model where she puts the pregnant mouse in a hypoxic chamber, and the pups come out small, presumably because of the hypoxia. We obtained placentas from Amanda and we measured SPINT1 levels in those placentas and those mice that were rendered hypoxic with small babies to SPINT1 levels in the placenta were lower. This adds even more scientific weight that SPINT1 may be associated with true placental insufficiency.
We next moved back into the clinic, and we looked at whether circulating SPINT1 levels were low many months proceeding birth of a small for gestational age baby, because this adds to biological plausibility that it may be a driver of placental insufficiency rather than being a bystander, which just happens to decrease.
We looked at our blood samples from our FLAG cohort of circulating SPINT levels at 28 weeks gestation, and indeed we found that those where we took bloods at 28 weeks and were destined to be born under the 10th centile SPINT levels as early as 28 weeks were ready lower compared to women who we took bloods at 28 weeks, but they ended up birthing a normal baby. And most of these babies were delivered at term. SPINT levels were already lower some many weeks before the onset of evident clinical disease, which was very exciting to us.
Next, we set out to validate our samples in other cohorts. So far, we’ve just shown you most of our data in Melbourne, and it’s very important to validate biomarkers in other populations in other settings. We went all the way across the other side of the world and we contacted our friend, Jenny Myers in Manchester, and she has the very impressive MAViS cohort. The MAViS cohort is a clinic which looks after very high-risk pregnancies. Pregnancies with significant vascular disease, such as chronic high blood pressure in the past, or type 1 diabetes, and they have a very high risk of fetal growth restriction and preeclampsia.
We found indeed in blood samples obtained in the MAViS cohort at 26 to 32 weeks gestation in the UK, circulating SPINT levels were indeed lower in those destined to birth the baby between the 5th to 10th centile and under the 5th centile compared to those with babies over the 10th centile.
After confirming this in the UK, we then moved over to Singapore and we approached our friends who run the impressive GUSTO cohort. This is 1300 participants where bloods were taken at 26 weeks, and there is very detailed follow-up of the pediatric outcomes from the GUSTO cohort. Again, we found in this group of almost 900 participants that if you had a growth restricted baby, the SPINT levels were low at 26 weeks compared to controls.
We next sort to see whether SPINT levels maybe even depressed at 20 weeks of gestation. Even early. In fact, halfway through the pregnancy, which is a tall ask, but we thought that we would have a look.
We also went to another country to at times chilly New Zealand and we approached the group who runs the impressive SCOPE cohort. They had 2000 samples and we measured circulating SPINT levels at 20 weeks gestation. This is halfway through the pregnancy, and we correlated whether the babies ended up being birth small many at term gestation. This is many, many months away to see where the SPINT levels were already low this early pregnancy.
Indeed we confirm that SPINT levels were lower in those who ended up birthing a small for gestational age baby compared to controls. We found that SPINT1 levels were lower in those destined to birth a baby under the third centile. This very much provides strong evidence that low SPINT1 may occur very early in pregnancy, and very much provides evidence that SPINT1 may have a role in causing placental dysfunction.
This is all very well to show SPINT1 is low, but we need to try and develop a diagnostic test which may be applicable to the clinic. Here, I want to show you a very exciting four tier risk model, which you can stratify the entire population according to their risk of having a low birth weight baby. We’re going back to the samples at 36 weeks gestation.
You can see that SPINT levels– You can split it into four tiers of risk. Let me just walk you through this slide. We’ll go with tier one, which is the high-risk group, and you have a particular SPINT1 cutoff, which represents the lowest levels. 7.1% of the entire population will fall into this group. If you screened everyone at 36 weeks, you’ll pick this. 7.1% of the population will fall in this group.
This is a group which arguably have a very high risk of birthing a small baby where I would suspect many clinicians would agree that this group would merit close monitoring and time birth perhaps at 38, 39 weeks gestation. Because of having approximately 3% chance of having a baby under the third centile, it goes up over fourfold to 14.1%. They have a 19.7% risk of having a baby under the fifth centile, they have almost tripling of the risks of having a baby under the 10th centile, and a doubling of the risk under the 20th centile. In fact, if you’re following this cohort, you almost have a half chance of having a baby under the 20th centile.
For 57% of the population, this test is unable to split your risk. For 27% of the population, it almost halves the risk of having a small baby, and arguably in this cohort, you may be more relaxed and allow women to resist intervention and await spontaneous birth safely. In 9.1%, there is a group with extraordinarily low risks of having a small baby. This is a group that you might arguably be particularly relaxed about.
This is using SPINT1. It’s an exciting approach, potentially, where you can use SPINT1 to triage the risk of having a small baby. You can identify those at increased risk and deliver them in time, but conversely, you can identify an important cohort where you will be able to abstain from medical intervention.
From our work so far, we conclude that low circulating SPINT1 may be a biomarker of poor placental function and growth restriction. It’s possibly the best published marker of placental insufficiency identified single marker.
Are we there yet? We don’t think so. We’re still on the hunt and this is the test that we aspire to develop. We want to identify a test with a 60% sensitivity, meaning it detects 60% of small babies. That’s because the ultrasound tests pick 57% and we would therefore find something which is simpler to administer, hopefully cheaper than lots of ultrasound. Will have less variability because it’s not operating dependent and will pick 60% of small babies. We’re not quite there yet. SPINT1 picks 45%. It’s way better than PLGF, but it’s not quite at that magic mark yet. We would try to identify a test which if it says it’s positive, there’s at least a 30% chance that the baby really is small. The ultrasound currently has a 35% positive predictive value and that’s in the best.
Our next step is to try and hit that level of a useful diagnostic, which we think we are tantalizingly close, is we’re still screening lots of blood markers. We’re going to combine SPINT1 and other markers we identify with ultrasound to see whether we can increase that positive predictive value, the accuracy over 35% and to see whether we can improve the sensitivity of ultrasound over 57%. Of course, we’ll look to combine maternal characteristics. We’re throwing the kitchen sink at this thing.
Combining biomarkers may have merit, and this is just early indicative data. This is under the fifth centile. SPINT alone had a sensitivity, can pick up 55.6%, when we add it with PLGF and two other markers, it goes to 63% and goes up a little bit, maybe not huge amount, but it is at least proof of concept that biomarker combinations just might outperform single biomarkers.
Other steps in the hunt is we need to look at larger cohorts and to discover and validate tests. We are still collecting thousands, and we have a great link with a machine learning expert or someone who’s very good and develops algorithms for artificial intelligence, and Polonia is helping us to see whether artificial intelligence will help from traditional algorithms to pick small.
Stillbirth is a devastating outcome. We are working like many other researchers funded by the Star Legacy Foundation to try and identify ways to decrease the devastating impact of a stillbirth. We are hopeful that we are close.
I’d like to acknowledge all that have made this possible. My friend, Professor Sue Walker, the scientific lead, Associate Professor Tu’uhevaha Kaitu’u-Lino, and our statistician Richard Hiscock. Thank you very much for your attention. Thank you so much to Star Legacy for this opportunity to present, and I wish you well for the rest of the conference. Thank you.
Please feel free to ask questions of the presenter. We will obtain their answers/comments and provide them here as received.