Predicting Perinatal Depression: The Future of Maternal Mental Health, a conversation with Dr. Sheila Shanmugan

Perinatal mood disorders represent one of the most common complications of pregnancy--1 in 5 women will experience this during or after their pregnancy; a sizeable fraction of these women experienced some kind of mood disorder prior to pregnancy--but currently it's hard to predict how pregnancy will mix with this history. My guest today shares with us what the future of pregnancy could look like; using advanced scanning technology, we may be able offer a much more accurate estimate of the chances of experiencing depression in pregnancy based on a look at a mom's brain structure​.

You can find Towards Personalized Clinical Interventions for Perinatal Depression: Leveraging Precision Functional Mapping here: https://www.science.org/doi/10.1126/sciadv.adt8104

You can find Dr. Sheila Shanmugan's (MD, PhD) work here, at the Perelman School of Medicine, University of Pennsylvania: https://www.med.upenn.edu/apps/faculty/index.php/g275/p8364165 

Charles (Chuck) Lynch et al 2024 Nature paper on mapping depression: https://pmc.ncbi.nlm.nih.gov/articles/PMC11410656/#Abs1

Dr. Laura Pritschet's episode on the show: https://podcasts.apple.com/us/podcast/its-not-mommy-brain-how-hormones-during-pregnancy-prepare/id1779600854?i=1000679984230

Transcript

[00:00:00] Perinatal mood disorders represent one of the most common complications of pregnancy. One in five women are estimated to experience this during or after their pregnancy. A sizable fraction of these women experienced some kind of mood disorder prior to pregnancy, but currently it's hard to predict how pregnancy will mix with this history.

My guest today shares with us what the future of pregnancy could look like. Using advanced scanning technology, we may be able to offer a much more. more accurate estimate of the chances of experiencing depression in and after pregnancy based on a look at a mom's brain structure. Welcome to Making Sense of Pregnancy.

This show is a new pregnancy reference. I'm finding and talking with experts doing cutting edge work to better understand what we do and don't know about pregnancy and what you can do [00:01:00] to better understand your own experience. Each week, I'll be talking to scientists, doctors, and researchers who are trying to uncover the many mysteries that still exist in reproduction, giving you the most current evidence based way to approach this enormous transition in your life.

I hope it will become your go to source for how to make your pregnancy better. Please enjoy my conversation with Dr. Sheila Shanmugan.  

Today we're lucky to have Dr. Sheila Shanmugan on the show. She's an assistant professor of psychiatry, OB GYN and radiology, it's a lot, and director of research for Penn Center for Women's Behavioral Wellness. Dr. Shanmugan, thanks so much for coming on the show.

Thank you so much for having me, Dr. Kamenecka. 

 So when I was looking up some of your work, , I'm going to quote here, you're described as, someone who's doing a pioneering effort toward personalized medicine and maternal mental health. 

Oh, wow.

Thank 

you. So nice to be a pioneer. Nice to talk to you. So I'm excited to get into this because one of the many things [00:02:00] you've published is a recent article in the journal Science Advances in which you talk about using fMRI technology to look at changes in women's brains during pregnancy to ferret out signs of depression. I thought this was one of the coolest things I've seen in a long time.

 But before we jump into this cool thing, let's talk about how we predict the onset of perinatal depression right now, without the use of fMRI. What are doctors doing currently to try to assess whether depression is likely in pregnancy?

Yeah, so that's a really interesting question. And I think it depends on if you're thinking about it from a research perspective or a clinical perspective.

 I was coming at this from a clinical perspective, and a lot of the research questions that my lab focuses on are really inspired by what we've seen is missing from the clinic. And what are the tools that I wish I had as a reproductive psychiatrist that would allow me to provide better care for my patients?[00:03:00] 

A really common clinical encounter for us is a preconception counseling encounter, and so that's where patients who are planning to become pregnant, , come in and talk to us about their plan for managing their medications and depressive symptoms during pregnancy. And so right now, what we basically do is we get some additional information, try to figure out what are the conditions the medications were started for.

weigh the risks of continuing these medications versus the risks of untreated depression, and try to figure out what are the risks, what is the risk of them decompensating? And so that's where the really big question comes in, because that's going to be different for each patient. And there's a little bit that we can tell.

So a little bit is based on a patient's history. Do they have some sort of mood sensitivity to hormonal fluctuations in the past? Do they have a family history? Do they have a history of trauma? What are the psychosocial stressors going on? But ultimately, while we have these indicators, it's a subjective assessment.

So we don't really have a quantitative marker for a given person on what's their risk of [00:04:00] developing perinatal depression. So that's what we're really trying to fill in the gap for here. 

So that, that actually speaks to my next question, which is , how accurate, and do we even know how accurate the current method is?

 Is anyone keeping track of, this is how we're doing it this way, and these are the number of cases we miss, and here's how many hits we get. 

 As far as I'm aware, someone isn't tracking the overall assessment of, here's how this is. How a reproductive psychiatrist assesses perinatal depression, and here's the absolute truth.

 I'm not sure how exactly we would come to that, conclusion. There are definitely studies that look at, screening methods. And here is what we're doing for screening. Here are the people we're missing. Here's how we can improve screening. But then once they come to psychiatry, , I don't think we're really quite keeping track of, we think this person is going to go on to develop perinatal depression.

Oh, they didn't. Or we think this person is going to be okay. Oh, they actually ended up [00:05:00] decompensating. I don't think someone is actually keeping track of that right now. 

Okay. Well, that makes sense. I understand that it is a judgment call and there are a lot of things in medicine that are judgment calls, but I guess if I'm the patient and we're trying to balance the cost of using medication in pregnancy versus the cost of, unmedicated, it.

depression, ideally I would have some kind of a sense of like the false positive rate or something like that to know how reassured I am , by the judgment call essentially., 

it's, it is more of an expert assessment, but it is subjective. And so that's where some of the uncertainty comes in. And so we can't tell someone the same way we would, say you were going to your primary care doctor, and they're monitoring you with blood draws to see if you have anemia, and they check your hemoglobin.

It's not the same, unfortunately. There's not something you like, Yes, here, your brain says that you have this, you don't have this. It's unfortunately, , Something we're still trying to figure [00:06:00] out, but that's where we hope research can fill the gap. There are excellent clinicians, and clinicians have decades of experience and training who are able to get the nuances of a patient's history.

And so, I think we do a good job. But there's definitely room for improvement.  

 The reason I ask is because when we're going to talk about the, your innovation, it seems like one of the things that could come out of that is a much more well defined sense of is this technology picking up something like a false negative rate or false positive rate, which I think is amazing in psychiatry.

That's what we're hoping for. 

Yeah. So that's amazing. So let's talk about your 2025 paper in Science Advances, where you talk about a personalized method to try to predict perinatal depression using precision functional mapping. So can you talk about what that is and how it's used? 

Yeah. So precision functional mapping is a newer method, , related [00:07:00] to functional brain imaging.

So Functional brain imaging or fMRI has been really important in uncovering some of the basic ways that the brain functions and what are some of the things that go wrong in certain disorders, but a major part of what's limited the clinical translation of this is some of the analytical techniques.

So you may have seen papers a few years ago or articles in the media about how fMRI, , isn't reliable unless you have thousands and thousands of subjects, , or participants and, , part of the reason for that is because of, , the way the data is analyzed. And so in traditional fMRI studies, You assume that functional networks are in the same anatomic location in all individuals, but more recently we have a lot of data that's come out that says that that's actually not quite true and the spatial layout of functional networks.

is actually pretty [00:08:00] different in individuals. 

When you're talking about functional networks, I'm trying to picture this. Many of us have seen one of those pictures where they have the brain broken into little regions, and this is responsible for these functions, and this is responsible for those functions.

Is that what a functional network is?

.

Yeah, so a functional network are basically physical regions of the brain that work together for a shared purpose. So it's not necessarily defined by where in the brain it is , but what are that brain region doing? And so what are the brain regions that are working together towards , certain cognitive functions or towards certain emotional functions.

And so regions that work together with a similar purpose are part of the same network. 

Okay. And so you're saying if you scanned your brain and my brain, they wouldn't look identical in terms of where those things are organized. 

Likely not. They would look similar, but there are going to be individual differences.

So you can kind of think about it like faces. And so everyone has a face [00:09:00] and they have the same features. So eyebrows, eyes, nose, and a mouth, but the actual size, shape, and exact location of the features vary between people. 

Okay. And I'm assuming the old fMRI studies are based on averaging across everyone, even though everyone's slightly different.

Yes. 

Okay, and that averaging is what's the issue? 

Part of what's the issue. 

Okay. Okay, so, this precision functional mapping is, , looking at an individual as opposed to averaging across individuals. 

Exactly. 

Okay, so it's looking at an individual over time. 

So you could do an individual over time or an individual at a certain point in time, but collect a lot of data at that point in time.

 It's really about meeting enough data to be able to define it at the individual level. So that's in part, why some studies haven't in the past because the amount of data that you get per individual is shorter. But for precision functional mapping, you get longer sequences or more data and other ways.

And using that additional data [00:10:00] is really important in being able to differentiate, what is the unique functional architecture, , functional network architecture in that person. 

Okay. And there's some model somewhere of how depression changes that functional map,   

Sort of. I think it's still being worked out. There have definitely been studies that look at, how are these networks different in people with depression, people without depression. There was, of course, that landmark paper, the, Chuck Lynch paper that came out, , I think last year now, that, found salience network or frontal striatal salience network expansion in individuals with major depressive disorders.

So there are definitely studies that are looking at, , how are these networks different in people with certain psychiatric disorders, but I don't think anyone's actually done that with perinatal depression. 

So just to give a little bit of detail about the Chuck Lynch study, it was published in Nature in 2024 and it also used precision functional mapping and deep sampling, [00:11:00] basically repeated looks at the same person or a long dense scan, and suggests that functional areas of the brain look different in people with depression.

There's one area in particular that has to do with reward processing, among other things, that is larger and is hogging some of the brain real estate from other areas. And this difference predates depression, so maybe it's a risk factor or a possible biomarker. And similar to what Dr. Shanmugan is talking about, you can imagine these scans being used to personalize treatment.

. Okay. To take a step back here, even to be able to look at someone's brain and understand whether there's a mood disorder because of a change in the shape of the brain or is it changes or is it like relative to someone else? 

So now you're getting into interesting questions. So there's two questions at play here, right? What does what your brain look like tell you about your risk for a certain psychiatric condition? And then how [00:12:00] does one person's brain change over a certain period of time and how does that impact risk?

And so those are both really important questions. So when you can think about it in the perinatal period, it's What does someone's brain look like preconception tell us about their future risk of developing perinatal depression? And also, how does their brain changes during pregnancy, how does that trajectory of brain change, what does that tell us about their individual risk of developing perinatal depression?

 Do we have answers to either of those? No. Okay, but this feels like this is moving. Psychiatry closer to the rest of medicine that can take specific, measurements of, let's say your liver, if you're a nephrologist, and, and figure out what the problem is. Whereas in the past, it seems like psychiatry has not had access to a specific part of your body that they can use to define whether you have something or you don't have something.

Exactly. Yeah. So right now, as I mentioned, that preconception counseling encounter, we do our [00:13:00] best based on an. A certain patient's individual history and risk factors to give a subjective assessment of do we think this person is going to decompensate during pregnancy or not? And that weighs into our overall recommendation of whether they should be on medication, not on medication, but.

We're hoping that the studies that we're planning are the first step in developing a quantitative biomarker that can eventually enter into the preconception counseling evaluation. So the plan would be, if someone tried to, sorry, I didn't mean to interrupt you. 

I was just going to say pioneer is too limited a word, you're like Magellan, this is amazing.

I mean, this will change psychiatry, right? This will fundamentally change. 

We really hope so. When I talk about it, I hope I'm not overselling, but I really do think this will fundamentally change the way reproductive psychiatrists practice medicine. 

I also think, 

but I 

also think as a patient, you will feel differently if you [00:14:00] have physical evidence of what's going on.

So it's not your fault. And it's not that you're too weak or, you know, a lot of people take on. something like depression as, something that's their fault I don't know. It's just, it's different than a broken leg. A broken leg is a broken leg, right?

 But this is making the brain more like that model where you can show physical evidence for what people are saying they feel, which is amazing. , and I feel like you're doing it two orders of difficulty because we know that pregnancy changes the brain. So now you're looking for two things simultaneously, right?

Yes. Do we think that, , the brain changes induced by pregnancy are different than the brain changes induced by depression? So 

we don't actually know that yet. And so the answer is we don't know. What we do know is that there is spatial overlap between the brain changes that occur during pregnancy and depression circuitry.

And so our hypothesis is that. [00:15:00] Because of this, in part because of this spatial overlap, the brain changes that are occurring during pregnancy, when they go wrong in some way, like may not follow a normal trajectory, that may predispose that individual to developing perinatal depression. 

 And so your idea is in these preconception visits, you're taking a mapping and you will know by the size of a certain functional area, whether that person is at higher risk.

So I think the first steps are delineating what you said in terms of in the preconception counseling encounter or in the preconception phase. Start to understand what are the brain changes in people who go on to develop, or differences in people who go on to developing a perinatal depression versus not, and try to isolate, or delineate those changes, but I think what I foresee it looking like in the future is As this expands, we get more and more samples, tens [00:16:00] of thousands, hundreds of thousands, eventually developing machine learning algorithms that use a person's personalized functional networks from their preconception counseling encounter to, , predict overall risk of perinatal depression.

So not based on the size of one network or a couple networks, but a high dimensional, , multivariate measure that can go into these models and and use that overall measure to predict risk. 

 I feel like you had to have come up with this, like, yesterday. To be able to use AI in that way. 

I interviewed someone recently. Dr. Jeffrey Goldstein at Northwestern, who's part of a study using AI to help them identify issues with the placenta based on a really large data set of placental images that are linked to diagnoses. So it's a little different than you have with the brain because it's once the placenta is not in use, and what you want is what's going on in this person who's living right now., 

if I were a pregnant person and I got this, evaluation, how would that change what I did in pregnancy? 

Yeah..

There are two [00:17:00] places. First would be prior to pregnancy. In that preconception counseling encounter, patient comes in, they ask us, should I continue this medication during pregnancy? And. In the future, once we have all this developed and validated, , and generalizable, integrate into clinical practice in a way that , these patients would undergo functional imaging as part of their preconception counseling and valuations, , that gives a quantity that would give a quantitative measure of risk of decompensating and then.

Based on that risk, could, way, should I continue medication, or should I not continue medication? So, for example, if the risk of decompensating is high, The recommendation would be continue medication not just because, , all the benefits to mom, but we know that perinatal, untreated perinatal depression has many consequences for mom, developing [00:18:00] baby, course of the pregnancy.

 And so we want to try to limit overall exposures and this could help us limit the exposure to untreated depressive symptoms when, the risk of decompensation is high. But then it also helps us limit the exposure to medication when the risk of decompensation is low.

So, for example, if it tells us that someone is not likely to decompensate, then they probably wouldn't need to continue the medication during pregnancy. And so that's how it helps us both from the physician perspective as well as from the patient perspective. And then I think the other part of that question was, , how would it help someone during pregnancy? 

 Say the model tells us, okay, high risk of decompensation. We want to continue medication., Some people are on medications with either limited or unfavorable reproductive safety data. And, for those patients, right now it's kind of in a tough place because There [00:19:00] aren't a ton of options.

 But one option that does currently exist is trans cranial magnetic stimulation. So basically using magnets to try to stimulate parts of the brain, targeting depression circuitry and limiting medication exposure that way, right now, that treatment only 30 to 50 percent response rate. It has been shown to be safe during pregnancy.

And so it's something that we are, we think that by using a person's specific functional brain architecture, we can improve the remission rates of TMS. And so that gives us another, uh, potential way that this would help someone during pregnancy. 

 So I have a better sense of what the cost is of using medication that has not been cleared for pregnancy, the, cost to the fetus.

, it sounds like there is no cost from trans. It's cranial magnetic stimulation. 

, so there's [00:20:00] not medication exposure. , so it, of course there are potential risks of transcranial magnetic stimulation or TMS. If you're going to the, , end of the spectrum, what's like the most severe, you can think about seizures.

And of course, if someone seizes, that would have risks for 

the baby. 

However, uh, very rare. You would want to titrate dose to make sure they're not seizing. So not to say that there are no potential side effects, but, it does limit medication exposure for sure. 

What does that look like if someone was going to get that treatment?

What are you wearing? , electrodes on your head or how does that work? 

, so they use a coil to stimulate, certain parts of the brain. You're not wearing electrodes on your head, but it feels like kind of like a Because that, uh, 

but it's all external, right? You're not, it's all external. Yeah.

Okay. Okay. And then we talked for one second about the cost to fetal development of untreated depression. 

Yes. So [00:21:00] untreated depressive symptoms can have consequences for a mom, baby. Course of pregnancy in terms of baby's development, , small for gestational age, low birth weight in terms of the course of pregnancy, higher risk of things like, , preeclampsia, gestational hypertension.

, So these are things that are untreated depressive symptoms. 

So some chemical shift in mood is affecting the way the placenta is feeding the baby? Wow. 

Yeah. So we think what's going on is that untreated depressive symptoms are causing changes in the neuroendocrine environment, and those changes in the neuroendocrine environment have downstream effects on the placenta.

That's so interesting. So they're sending signals to the placenta about how much food to get for the baby. 

Or how stressed you are pretty much. Yeah. 

Yeah. Oh my God. That's yeah that is a challenging trade off so yeah, this would make a huge difference both [00:22:00] in I think a pregnant person's sense of themself and kind of what to do. 

So talk a little bit about Having one of these kind of scans. What's that like? 

Yeah, so right now it would be about an hour. , we're still in the phases where we're trying to optimize. , what are the different methods we can use to make things shorter? , what is, , the right amount of data?

 What is the minimum amount? How much do you get by going longer? , Are there different things we can do in terms of sequences that can make the scans shorter? So we're still in the phase of, figuring out what. the eventual optimal protocol will look like. , right now in our research studies, , the scans for precision functional mapping take about, , 25 minutes out of that hour.

, and during that time a person is lying down in a scanner, , they are like in a MRI donut, would be like a typical MRI, head MRI scan. [00:23:00] 

Okay, okay, so I've had my share of MRIs which are not super pleasant But , but would I give up an hour to know if I was gonna get depression? I have 100 percent would but the another issue is that MRIs are expensive, right? And there's no other way to get that information at this point.

I mean From a clinical perspective or from a research perspective 

or for this to be, globally, adopted, it would have to be something that a lot of hospitals had. And, for sure in developed places, they all have MRIs, but. , I'm sure that MRI is running at all hours because it's expensive.

So there's a limit on the resources, right? , so there's no other way, clinically to, , get that kind of data. 

That's a good question. I think that there are, I think we don't know the answer yet. And so a lot of, we've talked about our hypotheses and what we. Are planning to do and what we hope we're going to get at the [00:24:00] end and show, but you bring up a good point that imaging is not the answer for everyone in all circumstances and.

In the course of this, we will also definitely be looking at what are the models we can develop based on existing data data from the electronic medical record, questionnaire, subjective report, are there ways that we can take the subjective information and develop a more quantitative marker?

And then the question becomes How do those compare like do the machine learning models that are built using the brain imaging at anything? Are they more accurate? How do they compare to those models that don't use the brain imaging and then figuring out what are the tradeoffs? And I think we don't have that data yet.

I think we, based on existing literature, believe that or hypothesize that in integrating brain imaging will lead to a more accurate model. , But it's not to say that [00:25:00] We would not still get, accurate translatable models without it. We just don't know yet. 

 The other thing I like about it now that I think of it is that I have interviewed many women about their pregnancies, many women who end up having, , postpartum depression and they do not want to answer those screens, honestly, either because they think there will be some consequence for, their ability to, to keep. Charge of their baby, or they think it means something about them personally as a mother, especially for first time mothers who have no sense of what that, transition will feel like. And so having some objective data, will hopefully, be a relief. 

Yeah, I hope so. , I think it will address the point you said, also make the decisions, maybe a little bit easier taking the burden off of the person who is experiencing the symptoms to make that decision to externalize it in a way.

As physicians, our role is to [00:26:00] provide and. Informed discussion and make sure the patient has all the information they have to make the decision, but ultimately it's the patient's decision on what they want to do, and I have seen patients struggle with that because, it's hard to get that information and then, you're left with that responsibility of, I wish I just had an answer and we're hoping that this can help.

Answer that question. 

 So where are we in this process? , are there centers around the USA where they're doing scans, in a, , experimental way so people could sign up for a scan for their own pregnancy? 

. We are doing that here at the University of Pennsylvania.

Okay. 

 We have a study going on right now that is doing, , brain imaging before pregnancy, or recruiting women, , people who are planning to become pregnant before they become pregnant, following them from the preconception time period. During pregnancy and through the postpartum with both imaging, questionnaires, clinical assessment, assessment of, serum, [00:27:00] saliva, , and hoping to use all of this information to try to answer those questions that we've posed in the paper.

So is Penn the only center doing this or is it? 

No, I'm, there are many people who are looking or. There are a few pioneers in this field. And so some people that I really look up to, for example, Emily Jacobs, Susanna Carmona, these labs are giants in the field and are really leading like the maternal brain project.

 The mother project. , they have published the, seminal papers in this area. Think one of them was also in the science advances are, , issue. Yeah. So, there are definitely people doing this. 

So if I were a pregnant person and I wanted to help with the science and figure out something interesting about my pregnancy, what would I do? How would I find out where I could be part of a clinical trial that's using MRI to study the brain in this way during pregnancy?  

, I think [00:28:00] it depends what area you're, what regions you live in. , 

am I going on the internet searching precision functional mapping, OB clinic? Or , what am I even looking for? 

I'm probably looking for MRI pregnancy research study.

Okay. Okay. Those are good. Search terms to look for. I couldn't be more excited. About your work. , it seems amazing. Thank you. And usually I ask, , how would you like to see, , maternity care change in the future? But I feel like this is probably how you'd like to see maternity care change.

Exactly. Exactly. . , amazing. Thank you so much for your work. Thanks for coming to, , share some of with us today. I am, I have fingers firmly crossed and I will be watching from the sideline cheering loudly. 

Thank you so much for having me and for shedding light on this field. This is a really important time for women's health, women's mental health, and we will take all of the exposure we can get to help spread the word on how important it is.

Thanks [00:29:00] again to Dr. Shanmugan for sharing some of her work with us today. After we finished taping our conversation, we talked a bit about the timeline for the introduction of this technology into common practice. And of course, it depends on funding, but Dr. Shanmugan was hopeful that with robust funding, we could see this in 10 years time.

For how dramatic a change this would be, the complete upheaval of how we look at, interact with, and think about depression and pregnancy Ten years doesn't sound that long. This effort is of course Helped by all the researchers doing some of this cutting edge work.

Dr. Shanmugan talked about her co authors on this paper in Science Advances Dr. Emily Badler and Dr. Laura Pritchett who has done other influential work on deciphering some of the massive changes in the brain induced by pregnancy And, who I interviewed in December, I'll link to that episode in the show notes.

Although this technology is not available to everyone right now, I think one takeaway that is, [00:30:00] is that depression has a physical manifestation in the brain. Perinatal depression is not evidence that you don't like being a mother, or you aren't good at it. It doesn't reflect anything about your strength.

There is physical evidence for what you feel. not so different from a broken leg. To the degree that it's not about attitude and not in your direct control, it may also make it easier to get help if you run into this very common issue. Thanks for listening. If you like this episode, please share it with friends.

We'll be back next week with more cutting edge research.