2024 Fetal Care Center Frontiers in Fetal Neurology Day 2 - Dr. Usha Nagaraj

Um, and so now we will move on to the next part of our presentation. So it's a pleasure to have Doctor Usha Nagaraj from the Division of Neuroradiology here at Cincinnati Children's, um, who will talk about the imaging findings with fetal opioid exposure. Um, I would like to thank Doctor Bena Tayson, uh, for the, and the rest of my colleagues at the Cincinnati Fetal Care Center for the opportunity to speak today and, um, share some of, uh, what we have learned over the past year looking at MRI of the fetal brain and opioid-exposed fetuses. This work was made possible by multiple funding sources within our department and a grant by the NIH, the Impact Trial. Um, the PIs in that study were Stephanie Murhar and Jennifer Vanist. So the stem of the interest in this work is the opioid epidemic is a big problem. Um, it continues to have a profound effect on neonates born in the United States, with 1 infant born with prenatal opioid exposure every 15 minutes. Opioid-exposed children tend to have lower school achievement and higher rates of severity of behavioral problems. However, the underlying neurocognitive mechanisms for these poor outcomes are unknown. However, fetal MRI shows great promise as a tool to evaluate the developing brain in vivo and may potentially help us answer some of those questions. So in the next 15 minutes or so, I'm just going to review some of the papers that we have published looking at this um cohort of patients, and I will start with the, the first paper which we titled MRI Findings in 3rd trimester, Opioid exposed fetuses, and this was um with a focus on brain measurements. So, I am a clinical fetal imager and pediatric neuroradiologist. So my interests mainly lie in things that I see on clinical imaging, things that I report in routine clinical practice. So my idea for this portion was just to compare the two. biometric measurements, many of which we routine, we routinely report um on clinical fetal MRIs, um, in addition to pregnancy-related assessments that we also routinely report and just see if there were differences between opioid-exposed fetuses and normal controls. So this was a prospective IRB approved multi-institutional study, the IMPAC Study. Um, patients were recruited at 3 sites, Cincinnati Children's Hospital, uh, University of North Carolina, and University of Arkansas Medical Center. Uh, Boston Children's Hospital was also involved in the study as a site for data analysis. They have a very, um, robust and well-developed, um, Neuroimaging program. Uh, patients, um, enrolled had to be at least 26 weeks gestational age at the time of imaging, and patients were divided into two groups, um, those with opioid exposure and unexposed controls. We did a total of 14 2D, um, measurements of the fetal brain in 4 derived indices, and we just compared the two groups using multivariable linear regression models, and we controlled for gestational age, fetal sex, and nicotine exposure. And we also reported some additional pregnancy related findings. So for our imaging protocol, all of our patients were scanned in our, our, all the patients recruited here in the research MRI scanner in the Schubert Research Clinic on a 3 Tesla magnet, and we required these images in 3 planes using 2 millimeter thick slices. This is not our typical clinical protocol. These um runs here through the brain took about 4 minutes each to obtain, which would be challenging to implement clinically. But we measured the brains in the same manner that we do in clinical practice. Here's some examples of the measurements. We additionally did a manual tracing of the vermis which we needed an external um computer software to perform. So for our results, um, we had a total of 65 women enrolled, 28 opioid-exposed fetuses and 37 unexposed fetuses. Um, so some of the things we noticed is, interestingly enough, breach presentation was more common in the opioid-exposed fetuses, 21% versus 3%, and a subjective assessment of amniotic fluid was increased in the opioid-exposed fetuses compared to the unexposed fetuses. However, um, fetal motion, um, which again was a subjective assessment, cervical length on fetal MRI and the deepest vertical vertical pocket of amniotic fluid were not significantly different between these two groups. This is a table, um, summarizing the demographics of the patients enrolled. Um, one thing you'll notice is that the nicotine exposure was, um, significantly higher in the opioid-exposed fetuses, and we did control for that in our analysis. Um, the other exposures were also statistically significant, and these were very heterogeneous in nature, ranged from alcohol and marijuana use to cocaine. Um, The amounts were very difficult to obtain. We did not control for that in this analysis. Um, birth weight between the two groups was not significantly different between the two groups, and also not reported here, head circumference between the two groups at birth were not significantly different. So in the adjusted models, we found a total of 7 measurements that were significantly smaller in the opioid-exposed fetuses compared to the unexposed fetuses. One was the brain frontal occipital diameter. Um, this also included the brain biparietal diameter and the bone biparietal diameter, so a measure of cal calbareal size. Also, the transverse cerebellar diameter. Verms height and in AP measurement of the pons were significantly smaller in the opioid-exposed fetuses. So we concluded that opioid-exposed fetuses have increased incidence of breach presentation and increased amniotic fluid that are unexposed fetuses. And the reasons for that are not entirely clear. Um, breach presentation might suggest, uh, abnormality in fetal movements. Our subjective assessment of increased amniotic fluid is unclear and may be related to fetal swallowing. Um, second, we found 2D, the multiple 2D brain measurements were smaller in the opioid-exposed fetuses compared to unexposed fetuses, and certainly the smaller brains fit with theories, um, around impact on neuronal proliferation by the, um, opioid exposure, though more work in this has to be done. So, uh, we have this great cohort of patients in the study which lent itself to other analyses. So here at Cincinnati Children's Hospital through the Imaging Research Center, we have, um, people who are very interested in diffusion tensor imaging or DTI. Weihong Yuan is, um, faculty at the Imaging Research Center here and Jonathan Dudley is research associate. And both of them wanted an opportunity to examine DTI in the fetus and um we thought that this would be a great cohort of patients to do it on given that we were doing it in the research setting. So, Some basics about diffusion, and I'll preface this by saying this, I am not an MRI physicist, but I can tell you what I do understand. And that's diffusion-weighted imaging is a sequence on MRI that allows for the evaluation of the movement of water molecules in vivo. So we have our routine DWI or diffusion-weighted imaging where um three gradient directions are applied to attain the average diffusivity. And then you have DTI, which is a type of diffusion-weighted imaging where you get 6 or more gradient directions, um, applied, which results in higher quality data. This can allow for the creation of tractography, which are anatomic representation of the white matter tracts in the brain, and it can also allow us to calculate, um, Numeric data including fractional anisotropy values or FA, which is a measure of the strength of directionality of the water molecules, and then the mean diffusivity or MDE, which is the strength of water movement. Now, I will say this, um, FA and MD values, well, for one, many of you may be familiar with literature describing these as potential biomarkers for Many, many different, um, diseases, um. But, um, the, the values need to be interpreted with caution. Um, they can vary, um, based on scanner hardware and acquisition parameters. So fortunately, um, working with, um, people like Doctor Yuan and Dudley make work like this feasible and more, um, robust. So, diffusion tensor imaging has been looked at in the brains of children with prenatal opioid exposures and other prenatal illicit drug exposures and have shown some microstructural differences in their white matter. Historically, DTI was not possible in the fetus due to fetal motion, but, uh, now due to slice to volume registration techniques which helped to compensate for fetal motion, it, it can now be done in a robust manner. It was actually used to create a detailed fetal DTI atlas by the Boston Children's Group. So the purpose of our study here was just to evaluate differences in DTI parameters in the brain between the opioid-exposed fetuses and unexposed controls. So we use the same patient cohort, um. The patients that had DTI performed, and these were all of course scanned on a 3 Tesla magnet and for our DTI protocol, we use 20 non-colinear diffusion weighted directions with a B value of 800. And the data processing was performed using SVR slice to volume registration. Now this is a post-processing technique which um automatically removes the motion corrupted slices and realigns the non-motion corrupted slices so that they're anatomically aligned. And then we used combat harmonization, um, which is a statistical correction strategy that minimizes the inter-center effects, um, In DTI resulting from scanner differences while um preserving the physiologic features. Uh, so this is just an example of the DTI images that I got from one patient just right off the scanner. So the first, the first image here is our T2 weighted image, um, which is our anatomic image. And then the middle image is our isotropic DTI or diffusion-weighted image of the brain. And then the third image is our color FA or fractional anisotropy map. And this is Without any post-processing, it just happened to be one of our best, um, studies, but the color coding in this map, um, is indicative of the directionality of the white matter tracks for those who aren't familiar, and we can see the splenium of the corpus callosum and the genu here and we can actually see the cortical spinal tracts on this one. But needless to say, most of our fetal DTI does not look like this. So this is where SVR came in and this was courtesy of Doctor Dudley. So here's an example of a patient in our study, um, where we can see first the T2 weighted images, the B0 images, and then the third column is our standard correction where we can see all these blotches of color. So, and that's the result of motion corruption. And then with the slice to volume uh registration, that's the last column here, we can see better delineation of the white matter tracks. The top images point to those blue corticospinal tracts, and the bottom is pointing to the cingulum. And this is a figure from the paper that we put out, um, showing how the um T2 weighted images were anatomically aligned to the DTI data to help for anatomic localization. Um, a total of 48 regions of interest were evaluated using a white matter atlas, and we looked at FA and MD values. So, We had a total of 28 patients, um, from, from Cincinnati Children's and Arkansas Medical Center that had DTI data. Um, we had to exclude, well, more than half of them due to excessive fetal motion, just illustrate some of the challenges we have with these advanced imaging techniques. So we, uh, in total, we had 15 controls and 11 opioid exposed differences. Of note, we did, um, Look at differences in fetal motion, and there was no statistically significant difference, and this was not a subjective assessment. This was a measure of the average framewise displacement in millimeters. So what, how much the, the image actually shifted, um. was measured. And there was no significant difference. So here's a summary of our results. So we, we found. Statistically significant higher FA values in multiple white matter regions, including the bilateral medial lemnisci, uh, the middle cerebellar peduncles. Um, And, and those were the ones I think, yes, those were the ones with statistically significant difference, and we saw some others with trending statistic um trending significance, um, also with higher FA values. So what does this mean? Well, DTI with SVR may reveal microstructural differences between opioid-exposed fetuses versus normal controls. And as far as what the specifics of what our data means is how is it different, it's really unclear at this point. Um, we think that opioid exposure may potentially cause inflammation affecting normal white matter development, but clearly further studies in a larger patient populations will be needed to better understand these findings. So now I'll move on to the, the 3rd paper. Um, so this, uh, was made possible by, um, the Boston Children's Group, um, uh, and it was very, uh, and I really enjoyed working with all these people and learning so much from them, but, um, they taught me about how, uh, these advanced fetal MRI imaging techniques can be used to evaluate the brain. So Ellen Grant, um, is one of the um co-I's in this project. She is the director of um fetal and neonatal um neuroimaging, um, at Boston Children's Hospital and Keho M is research faculty and, um, the And PI in the Developmental neuroimaging lab. And then Heu Jun-yun is the um research associate at Boston Children's and he did a, a lot of the legwork that made this work possible. So, What we learned, what I learned is that um advanced MR processing techniques um can allow us to uh find more detailed and precise information regarding um differences in brain volume, surface area, sulcle depth, um, cortical folding measured by mean curvature and gyrification index in a way, um, that can allow us to objectively assess these parameters. Um, and these metrics may have the potential to help us better understand the underlying causative mechanisms of opioid exposure. So, the purpose of this study was just to look at these morphologic features using these advanced processing, post-processing MRI techniques and compare the opioid exposed and the unexposed fetuses. Uh, so, in this study, same patient cohort, um, all of the MRIs that were performed in all the fetuses were sent to Boston Children's, and they created 3D, uh, volumetric images of the brain, um, using slice to volume registration technique. And they used a specialized pipeline to help them automatically segment the cortical plate in these fetal brains, so they could get data on brain volume, cortical plate volume, and surface area. They were also able to use the data to calculate, um, Information on the gyro cyclical pattern, including mean curvature and something called a gyrification index, which was a ratio between the surface area and the convex hull, which I guess is kind of like the shell around the brain and um it's such that the Increased, uh, gyrification index, um, indicates increased convolutional markings in the brain. Uh, so this was, uh, this was a slide illustrating that courtesy of Hu Yu Chun, but the first image here shows the raw fetal MRI image. So what, um, I would see at the workstation, if I were reading a fetal MRI, here's mom's cervix here, here's mom's bladder, and then here's our fetal brain here. And then after they did, um, after they got the images, they extracted the brain, um, and created the SVR image. They did a quality assessment and motion correction process, and then they aligned the images to allow for automatic tissue segmentation, and then the final step of this was cortical, uh was, was surface extraction to. Um, create the cortical surface model. Now, I will say all of the details of these, um, processes were quite challenging for me to understand, but it's, it looks very involved and very mathematical, and certainly this is something that takes hours per patient is, is what I, I was told. It's, it's very, um, labor intensive. Um, so, and so for our results, so we have the same 63 patients from all three sites. However, unfortunately, over half of them had to be excluded for inadequate image quality. Um, the fetal motion was really, um, challenging to work with, with the, um, uh, with the accurate fetal brain extraction. So we had at the end, 29 patients included, 14 opioid exposed and 15 unexposed, and in this analysis, we corrected for gestational age, fetal sex, maternal education, polysubstance use, high blood pressure, and the site at which the MRI was scanned. So, um, this table summarizes the findings. Essentially, all of the parameters assessed were were lower in the opioid-exposed group compared to the controls. This included total brain volume, cortical plate volume, surface area, sulcal depth, depth, mean curvature, and the gyrification index. And these are some statistical maps that Heujin put together, um, showing, uh, some regional, um, uh, preferences for these areas of, of, of, uh, change. So, uh, the first, uh, uh, the first set of images here, uh, this shows, so the red shows areas of decreased, um, Uh, surface area in the opioid exposed patients compared to the unexposed. And then this FDRQ value shows the areas of statistical significance, and we can see that there is a, um, regional preference for the perisylvian regions and central sulci. Uh, when we look at the same images for sulcle depth, um, the, uh, uh, the decreased sulcle depth is a little more widespread, but again, we do have some suggestion of regional preferences to the bilateral perisylvian, bilateral central sulci, and parieto-occipital features. So there may be some underlying, um, Um, functional, um, ideology for the uh Uh, resulting, um, anatomic disturbances. And then the mean curvature we shot saw no no statistically significant, um, regional differences. So what we concluded were that decreased uh volumes and gyration of the brain and fetuses with prenatal opioid exposure compared with unexposed controls using advanced fetal MRI processing techniques, and we think this is possibly related at least in part due to opioids causing impaired neurogenesis in the developing brain. However, given the complexity of human brain development, there are likely multiple underlying mechanisms affected. I would, I would like to acknowledge some of the people that made this work possible, especially, um, Doctor Stephanie Murhar, and I'm happy to take any questions at the break. Thank you so much.