Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors
Generic selectors
Exact matches only
Search in title
Search in content
Post Type Selectors

Epigenetic Gestational Age Prediction

Listen or watch on your favorite platforms

Determining a newborn’s due date traditionally relies on maternal reports of the last menstrual period and ultrasound scans.

These conventional approaches can lead to uncertainties, especially when it comes to identifying deviations from normal fetal development that could impact research into the effects of preterm or post-term births on newborns.

However, researchers, including Kristine Løkås Haftorn, have now developed a more precise method to ascertain newborns’ gestational age through analyzing DNA methylation patterns in blood samples, utilizing machine learning.

This is crucial because accurate knowledge of gestational age is fundamental for understanding the risks and implications of preterm and post-term births on infant health.

Moreover, the ability to accurately determine gestational age in utero could revolutionize prenatal care by providing deeper insights into fetal development, potentially allowing for earlier identification of developmental issues and more tailored interventions to support healthy pregnancies.

This breakthrough, driven by machine learning’s ability to sift through and interpret complex epigenetic information, underscores the potential of combining technology with biology to enhance our understanding of human development.

In this week’s Everything Epigenetics podcast, I speak with Kristine about epigenetic gestational age prediction, how we can use gestational age clocks to look at developmental timing and how this can improve pregnancies, assisted reproductive technology (ART), and more.

Kristine is particularly interested in epigenetic patterns in newborns, how these patterns are linked to development in the fetus and child, and how they can be affected by various exposures during pregnancy.

In this podcast you’ll learn about:

– DNA methylation’s role in fetal development
– Gestational age and how is it linked to fetal development
– Predicting gestational age using epigenetics
– Why determining specific cell types responsible for an association between DNA methylation and a given phenotype important
– How Kristine is adjusting for cell type composition in her work
– What cell-type specific DNA methylation patterns are associated with gestational age
– Nucleated red blood cells
– Why Kristine believes nucleated red blood cells are the main cell type driving the DNAm-GA association
– The poor correlation observed between epigenetic age clocks for newborns and those for adults
– How we can use gestational age clocks to look at developmental timing and how this can improve pregnancies
– Assisted reproductive technology (ART)
– Differences in disease in ART babies and traditional birth babies
– Epigenome-wide association studies of ART
– Investigating CpGs on the X chromosome
– How Kristine’s research will affect ART protocols in the future

Kristine obtained her bachelor’s degree in molecular biology and biochemistry and then a master’s degree in molecular bioscience at the University of Oslo. For her master’s thesis, she worked in Vegard Wyllers group at Akershus University Hospital on regulation of gene expression in adolescents with chronic fatigue syndrome/myalgic encephalomyelitis (CFS/ME), focusing on microRNA and DNA methylation. After finishing her masters, Kristine worked as an advisor at the Norwegian Institute of Public Health (NIPH) where her main tasks were to arrange a Norwegian research conference on CFS/ME. After that, she got a PhD position at the Centre for Fertility and Health at NIPH. During her PhD, she has been working on the relationship between gestational age and DNA methylation in cord blood, focusing on prediction models and underlying biological mechanisms using data from the Norwegian Mother, Father and Child Survey (MoBa).


Hannah Went (00:00.514)
Welcome to the Everything Epigenetics podcast. Christine, thanks for being here today.

Kristine Løkås Haftorn (00:05.252)
Thank you, it’s an honor.

Hannah Went (00:07.662)
Yeah, I know this is the first time we’ve actually ever spoken, you know, face to face over the phone, whatever you’d like to call it. So we’d love to just hear a little bit more about your background. I gave the listeners, you know, a little introduction into your journey and what you’re studying now. But if you could talk a little bit more about, you know, your interest and how you ended up where you are today, we’d love to begin there.

Kristine Løkås Haftorn (00:32.204)
Yeah, well, I’ve always been very interested in nature in general and animals especially. And always been very eager to learn new things. I think that’s kind of why I ended up here in research. I don’t think when I was little, I don’t think it was very obvious that it was biology I was going into because I really I was interested in basically…

anything new that I could learn. But I think it was in high school in biology when we learned about biotechnology and genetics, it just really sparked an interest and I just found it so fascinating. So I ended up starting to study molecular biology. And during my bachelor’s, I just really fell in love with.

cells. And just, I don’t know, I’m just so amazed by how these small molecules and small signals form this complicated organism and how these tiny little things inside us can create different phenotypes. And yeah, I just find that so interesting. So when I kind of got the

Hannah Went (01:28.578)

Kristine Løkås Haftorn (01:56.52)
research in my master’s project. I just found out that I can’t really do anything else. This is what I’m supposed to do. Because I just really love learning new things, reading about research and the creative side of just coming up with new ideas and projects and trying to find out new things.

Hannah Went (02:20.834)
Sure. Yeah, definitely you’re calling. And I think epigenetics is a great place to start because there’s so many things that we can learn. So I know in your PhD in particular, you’ve been working on the relationship between gestational age and DNA methylation in cord blood, focusing on prediction models and underlying biological mechanisms. So what really made you start to look into epigenetics and DNA methylation within your work?

Kristine Løkås Haftorn (02:47.588)
So it was, I mean, I started getting interested in epigenetics when I was doing my masters. When I did my, I looked at regulation of gene expression in adolescents with chronic fatigue syndrome and that was kind of when I started to get more interested in the different ways of regulating gene expression and learned more about DNA methylation.

Hannah Went (02:55.34)

Kristine Løkås Haftorn (03:18.028)
Starting my PhD, I really wanted to kind of continue and learn even more about epigenetics and DNA methylation. So when I started, the project was quite open and I was able to form it a bit out of my own interests. It was supposed to be about epigenetic aging, but then I kind of, because one of my supervisors had done some work on gestational age and

It’s a lot of work that is being done on aging, but not that much on gestational age. So I just found that there was so much to explore. And so that’s kind of where, why I ended up with that. And also because I had the opportunity to use this amazing data set that I, that I kind of had in my group. So.

Hannah Went (03:54.689)

Hannah Went (04:08.298)
Yeah, and we’ll definitely be digging into that today. And why I reached out to you in the first place is because I saw that paper be published by you. So moving into that, it’s gonna be mainly focusing on the paper titled Nucleated Red Blood Cells Explained Most of the Association Between DNA Methylation and Gestational Age. So before we dive in, I wanna talk more big picture. What do we actually know about DNA methylation and its role in fetal development?

Kristine Løkås Haftorn (04:37.9)
Yeah, I mean, we know that a functional DNA methylation machinery is essential for normal mammalian development. And we know, for example, that mice with inactivated DNA methylation enzyme genes, if they don’t have, or these genes are inactivated, so they don’t have these enzymes that

These mice get severe developmental abnormalities and they die during the embryonic period. So it’s clear that it’s needed. And it’s a lot of work trying to get into why we need it and what specifically it does. And we do know that in the early development, we have these extensive genome-wide waves of methylation.

Hannah Went (05:18.626)

Kristine Løkås Haftorn (05:35.532)
reprogramming during embryogenesis. So there are two of these, one shortly after fertilization and the other is after germline cell specification. And during these events, most of the DNA methylation is removed and then reestablished. So we know that DNA methylation is implicated in a range of different processes. I mean, it’s been.

associated with gene expression, but also more particularly in development, we know that it’s important for genomic imprinting, where a gene is differentially expressed depending on whether they are inherited from the mother or the father. And then we also have X inactivation, where one of the X chromosomes in a female is inactivated. That’s also DNA methylation. And also…

DNA methylation is very important for repression of transposons, which are small pieces of DNA that can kind of jump from one location of the genome to another and can thus lead to mutations, and also repression of germline specific genes. And then we have the tissue specific regulation of gene expression, where we see that the different cell types that kind of develop throughout development will

get these very different DNA methylation patterns. So it’s clear that DNA methylation plays a role there as well.

Hannah Went (07:07.182)
Yeah, I think it’s crucial to our being kind of echoing what you just said. It’s happening in all of these different ways. So I appreciate you kind of going through those examples. And I’m curious myself to learn a little bit more as to how this applies to gestational age, because as you mentioned, we focus more on just the general epigenetic biological aging. There’s very limited research on the gestational age. So how are we actually defining gestational age? Can you just…

Kristine Løkås Haftorn (07:10.741)

Hannah Went (07:35.082)
maybe define that for our listeners so we’re all on the same page and then talked about how gestational age is actually linked to fetal development which is also very important.

Kristine Løkås Haftorn (07:43.68)
Yeah, sure. I mean, gestation is the period between conception and birth. So gestational age is the duration of pregnancy, which is usually then measured from the first day of a woman’s last menstrual period and to the current date. So that could be any point during pregnancy or at birth, where we call it gestational age at birth. And it’s one thing that is important to note there is that

when we use this definition, the estimated gestational age is actually two weeks longer than the actual time that has passed since conception, because there is this interval between the onset of the last menstruation to the actual ovulation and conception. So the ideal measure of gestational age would be from the day of conception to the date of delivery, but we don’t really…

have any methods that can measure or detect the actual conception with such certainty. So gestational age as opposed to chronological age can only be estimated approximately. I mean you know when you’re born but you don’t know when you’re conceived. So we only can estimate it usually by either this just looking at the last day of last menstrual period.

Hannah Went (08:53.742)

Kristine Løkås Haftorn (09:08.556)
a bit can make it a bit difficult because of course the length and regularity of menstrual cycles can vary. Contraceptive use can also change the length of the last menstrual period. And also you can also use of course ultrasound which has been more common the last decades where

you take some measurements with ultrasound and then you kind of compute or estimate the gestational age based on that. And that’s generally more accurate than using the last menstrual period. But then again, there are some problems because it assumes that all fetuses have an equal size at one particular time point. So differences in early growth can lead to bias in that estimation as well. So there’s no…

like perfect estimation of gestational age. And because the fetus is developing and growing throughout gestation, gestational age is closely linked to fetal development. So for example, preterm birth, which then would be a low gestational age at birth is associated with a range of health outcomes in the newborn and later in life, such as risk of several diseases, risk of death.

Hannah Went (10:05.811)

Kristine Løkås Haftorn (10:32.516)
developmental delay, cognitive and behavioral adversities, and gestational age at birth is often used clinically as an indication of developmental maturity.

Hannah Went (10:44.278)
Mm-hmm. Yeah, absolutely. Thank you for defining that for our listeners. And we know that, yeah, if you’re outside of the particular gestational age window that can extremely affect fetal development. Is that accurate to say? Yeah. Well, good, good. Thank you for, again, reviewing that definition. I wanted to make sure everyone is just on the same page here, because my next question I’m extremely interested in.

Can we actually predict gestational age using epigenetics? Is that what we’re trying to do here? I know you were kind of saying there’s no real way to measure a gestational age or each methodology, I guess, has its particular downside. So can we use epigenetics to measure it?

Kristine Løkås Haftorn (11:27.412)
Yeah, I mean, just like chronological age is closely associated with DNA methylation, gestational age is as well. So there has been developed a few gestational age clocks based on the same methodology as the age clock. So you use some statistical algorithms to pick out some CPGs that are highly predictive of gestational age. And

Hannah Went (11:33.774)

Kristine Løkås Haftorn (11:53.248)
I mean, we published a gestational age clock a couple of years ago, and that had an accuracy of three to four days, so it’s quite accurate. But of course, to train these models, these clocks, we need to have some sort of standard or something to kind of measure up to. So we use ultrasound measures or ultrasound estimations as kind of the golden standard.

seeing as these are not really necessarily that accurate, it’s kind of, yeah, we can’t get more accurate than what we’re actually using as a template for our clock.

Hannah Went (12:28.728)

Hannah Went (12:36.51)
Yeah, are there any other, are there any other, I guess, inputs that you would use for the clock besides the ultrasound? Or that’s the most accurate way? So it is a little bit hard to do that then.

Kristine Løkås Haftorn (12:47.437)
I mean…

Yeah, so I mean you could use LMP, but my supervisor actually published in 2016 when he published one of the first gestational age clocks. He looked at the, he tried to use LMP to make the clock and then ultrasound to make the clock and it got more accurate using ultrasound. So, but what we also did in the…

Hannah Went (13:12.142)
Great, gotcha.

Kristine Løkås Haftorn (13:17.816)
paper that we published two years ago, we actually used data from children born after IVF, so Assisted Reproductive Technologies. And we used, because then we knew the actual date that these embryos were inserted. So that’s kind of the closest you can get to knowing the actual date of conception. So that’s, we tried to make a clock using that measure and it wasn’t really…

Hannah Went (13:37.091)
Mm. Yeah.

Kristine Løkås Haftorn (13:46.056)
it was approximately the same accuracy. So it seems like we’re kind of as accurate as we can get with the clocks that we have now.

Hannah Went (13:53.322)
Yeah, that’s great to hear. And what you say was the other thing you were using besides the ultrasound, the LPN?

Kristine Løkås Haftorn (13:59.572)
Oh yeah, sorry. It’s the last menstrual period. So that’s the date of the last menstrual period.

Hannah Went (14:02.45)
Oh, okay, gotcha. Perfect, well that’s good to know though that you’re getting these clocks as accurate as possible, which I think is really important if you’re trying to measure the gestational age. And then would you say that the application of those clocks in particular are going to be used maybe in the future to estimate that gestational age and maybe predict for certain outcomes with those newborns then and maybe take more preventative approaches? Or what do you think those exact applications would be?

Kristine Løkås Haftorn (14:28.372)
I mean, it depends a bit. I think the gestational age clocks are quite early stage compared to the age clocks so far. And it depends on what you want to achieve if you want to get as accurate estimation as possible for the gestational age, or if you’re looking at, or if you’re trying to measure the biological maturity of the newborn. So those are.

to slightly different questions. So, and we still don’t really know what the clocks are measuring, what those CPGs are doing. And I think we need to kind of do a bit more research into that before we know how much we can, or what we can use those clocks for later, or in the future. But I think definitely they can be useful in…

maybe for a clinical evaluation of pre-terms, maybe it can tell us something more about the maturity of those newborns. So far we can only predict after birth, so it’s not like we can, it’s not like an ultrasound that you can do in the middle of pregnancy, but maybe in the future we will be able to do that as well, have some measurements during pregnancy, so.

Hannah Went (15:36.11)

Hannah Went (15:42.839)

Hannah Went (15:51.15)

Kristine Løkås Haftorn (15:55.46)
So I think definitely it’s a very exciting field to be in because it’s a lot of things to explore.

Hannah Went (15:56.078)

Hannah Went (16:03.602)
Yeah, yeah, I think it would be interesting even to, I didn’t even have that thought about being able to measure still during the actual pregnancy to maybe look at different outcomes. So it’d be interesting to evaluate those gestational age clocks and figure out which CPGs are being used within those clocks and then what those CPGs are associated with, I think is what you’re hinting at in terms of learning a little bit more about what that can actually tell us. So I’ll be interested to keep following your work, Christine. So.

Kristine Løkås Haftorn (16:26.436)

Kristine Løkås Haftorn (16:31.781)

Hannah Went (16:32.746)
Going back just to the DNA methylation and its role in fetal development, why, you talk a lot about this in your paper, why is determining if specific cell types are responsible for an association between DNA methylation and a given phenotype important? And this is something that I’ve talked about. This is a recurring theme on my podcast. So why is it so important in your case and just when we’re looking at these markers in general?

Kristine Løkås Haftorn (16:57.312)
Yeah, I mean, DNA methylation, as I mentioned earlier, is very cell type specific. So that means the patterns of DNA methylation can be very different between different cell types. And different cell types usually have different functions and different processes going on. So if you find an association between DNA methylation and a phenotype, and if you’re like me and really want to understand what’s going on biologically,

Hannah Went (17:02.882)

Hannah Went (17:23.892)

Kristine Løkås Haftorn (17:26.412)
Determining which cell types these changes are happening in is a good step on the way, because then you can kind of go more into those processes and trying to understand what is actually going on.

Hannah Went (17:39.958)
Yeah, I think it’s kind of its own exploratory field within epigenetics itself, right, is controlling for those immune cell subsets and trying to do that. So at True Diagnostic, we’ve done a lot of work with separating those immune cell subsets and have a preprint where we’re kind of creating an immune cell composition deconvolution method for 12 different immune cell subsets. And I know the Buck Institute is working on that same information as well. Theirs is actually called the intran clock that they’re using.

Kristine Løkås Haftorn (17:46.068)
Yeah, for sure.

Hannah Went (18:08.158)
So what about you? How are you adjusting for cell type composition? I know there’s a lot of data normalization and pre-processing methods and pipelines. Which algorithms are you using?

Kristine Løkås Haftorn (18:19.488)
Yeah, so I mean, in this paper, we’ve done, we haven’t really adjusted for cell type composition in the typical way. Because, I mean, just to give a bit of background, I mean, when you analyze the tissue that consists of many cell types, you get this average of all the cell types in the tissue. And as you mentioned, we study cord blood and the

Hannah Went (18:29.889)

Kristine Løkås Haftorn (18:47.992)
The composition of cells in the cord blood can differ quite a lot between different individuals. So this, of course, then can introduce bias when you look for associations between DNA methylation and gestational age or whatever phenotype you want to look at, because the change that you see may just be due to the differences in cell type composition. And

Hannah Went (19:09.076)

Kristine Løkås Haftorn (19:11.784)
We are using cord blood that has been frozen and thawed without being treated in a particular way before freezing, so we’re not able to measure the actual composition of cells. And therefore, we use these statistical methods to estimate the cell type composition of each sample. That’s based on DNA methylation patterns that are shown to be very specific for the different cell

Hannah Went (19:30.242)

Kristine Løkås Haftorn (19:41.42)
method to estimate the composition. And then we get these proportions of each cell type in each sample that we can kind of include as variables in our model if we want to adjust for them. But what we’ve done here, because when just adjusting for these variables, you don’t really get any information about which cell types are driving the association. You just get.

Kristine Løkås Haftorn (20:10.7)
something that is, or kind of an adjustment. So what we did in this paper was to apply a different kind of statistical algorithm, which is called cell DMC, which basically looks for interactions between the cell type proportions and the phenotype, so gestational age in our case. And the key idea here is that if a DNA methylation change is happening with increasing gestational age in one specific cell type.

there should be a significant interaction between the proportion of that cell type and gestational age. So it’s a quite simple regression model where you just include this interaction part, and that makes us able to get cell or specific kind of significant CPGs for each cell type.

Hannah Went (20:48.536)

Hannah Went (21:04.654)
Sure, so to repeat what you just said for our listeners. So you’re, no, that’s great. You are looking at essentially just an association, right? So if you have a higher amount of one type of cell types that’s associated with gestational age, you may say, okay, well, the increase in this particular cell type may be responsible for gestational age. Is that an okay way to put that?

Kristine Løkås Haftorn (21:07.768)

Kristine Løkås Haftorn (21:29.56)
Yeah, but it’s kind of also we are trying to see if there is gestational age specific change in that specific cell type. So not just with the increase or decrease of cell type, that’s kind of what we use to infer those results. But we want to know.

if there are any changes that are happening specifically in those cell types and not in the other cell types. So, yeah.

Hannah Went (22:01.966)
Mm-hmm, understood. Gotcha, so yeah, what’d you see then? And I probably already gave this away by talking about the title of the paper, which I didn’t mean to do. But what cell type specific DNA methylation patterns then are you seeing associated with gestational age?

Kristine Løkås Haftorn (22:09.872)
I’m sorry.

Kristine Løkås Haftorn (22:18.424)
Yeah, I mean, we actually found significant DNA methylation associations with gestational agent in all the seven cell types that we included in our analysis. Some are overlapping between two or three cell types, but many are very specific to only one cell type. And what surprised us was that more than 2,000 CPGs, so almost 90% of the cell type-specific association that we found were in those nucleated red blood cells.

Hannah Went (22:28.204)

Kristine Løkås Haftorn (22:47.56)
even though this was not the cell type that was most abundant or varied most with gestational age.

Hannah Went (22:56.254)
Yeah. And why do you believe you saw that? Or I guess maybe do you want to define a nucleated blood cell as well? Or red blood cell for our listeners?

Kristine Løkås Haftorn (23:04.017)
Yeah, I mean the nucleated red blood cells. I mean I have had never heard of them before. I did this analysis to be honest. So I really had to dig into the literature and try to figure out what I was seeing because I wasn’t expecting this at all. But they are actually immature red blood cells that have not yet removed their nucleus because in healthy adults the red blood cells are enucleated so they remove their nucleus before they enter the circulation, the blood stream.

Hannah Went (23:08.088)

Hannah Went (23:13.833)

Hannah Went (23:17.73)

Kristine Løkås Haftorn (23:32.496)
But in the fetus, a proportion of these circulating red blood cells still contains a nucleus. And the presence of these nucleated red blood cells in the fetal and also newborn circulation is most likely due to the very high demand for red blood cells and oxygen in the kind of state in the womb.

Hannah Went (23:54.112)

Kristine Løkås Haftorn (23:59.41)
But they also have some regulatory functions of the immune system. So they have several different functions, but it’s most likely this oxygen demand that is most important and why they have these nucleated red blood cells.

Hannah Went (24:16.054)
Yeah, I’m glad you cleared that up because at first too when I was reading your paper, I was like, wait a second, what’s a nucleated red blood cell? All I know is that in adults, they’re not supposed to have DNA or have the nucleus. So I did some literature digging of my own, but I didn’t wanna get any feedback or comments from any of the listeners saying, wait a second, this is incorrect. So these nucleated red blood cells are immature red blood cells.

Kristine Løkås Haftorn (24:21.816)

Kristine Løkås Haftorn (24:32.633)

Kristine Løkås Haftorn (24:36.792)

Hannah Went (24:42.022)
We believe we see these and we believe that they’re the main type driving the DNA methylation gestational age association because in newborns these are needed for growth for the oxygen like you were just stating as well. So does that make a little bit more sense then once you did digging into the literature as to maybe the association that you saw?

Kristine Løkås Haftorn (25:04.04)
Yeah, yeah, for sure. And I mean, we, when we looked kind of into this, these functions of the nucleated red blood cells, and also the CPGs that we discovered, it really made a lot of sense biologically, because since they are, these cells are an important part of the red blood cell development. And red blood cell development and oxygen transport is crucial for fetal growth.

We also saw that these CPGs that we found, they were also in or near genes that are involved in red blood cell development. And one interesting process that showed up was the response to glucocorticoids. And glucocorticoids, they are most commonly known as stress hormones, but they are essential for a wide variety of biological processes. And they play…

important roles in pregnancy and normal fetal development as well, and regulation of red blood cell development. So it really kind of fit together this whole story and we also found CPGs in genes that are essential for the switch from fetal to adult hemoglobin, which is a process that occurs around birth because of the changing oxygen environment of the newborn.

So, yeah, it was very fun working with this because it’s usually you do analyses and even either they don’t work out or you get results that you don’t really kind of, yeah, they don’t really mean anything or you can’t really figure out what it is. But here it just kept on coming things that really fit into the picture. So that was so much fun working with this paper.

Hannah Went (26:49.454)
Correct. It seems like it just made sense, right? Even though you got an unexpected result, you do a little bit more digging, you find the mechanism of action, and you’re saying, okay, you know, these pieces of the puzzles are starting to fit together really well. So it makes for a really nice story. Christine, what else did you find from the paper? I know you said you looked at what? Seven different cell types. Some of them were a little bit overlapping. I guess nucleated red blood cell is what we saw the.

Kristine Løkås Haftorn (27:02.879)
Yeah, for sure.

Hannah Went (27:14.482)
again the main cell type driving this association between DNA methylation and gestational age. Were there any other large takeaways you wanted to mention from the paper?

Kristine Løkås Haftorn (27:23.2)
I mean, yeah, we didn’t really have enough time to dig into the other cell types as well. It was a large paper on its own. So that is something that I really want to do some further work on because I’m quite sure that some of the immune cells also play a role in gestational age. So, but what we found that was very, also very interesting.

Hannah Went (27:44.224)

Kristine Løkås Haftorn (27:53.144)
because these nucleated red blood cells, they decline rapidly and disappear from the circulation during the first few days after birth. So what we see is that when we look at gestational age clocks and adult age clocks, we see that there’s not really a lot of correlation between those. I mean, the CPGs are totally different.

when you try to predict gestational age with an adult age clock or vice versa, it doesn’t really work very well. So that was also kind of a hypothesis that was generated by this work that if these NRBCs are actually the primary drivers of this association that we see in gestational age, this can kind of explain why.

this poor way we see this poor correlation between epigenetic age clocks for newborns and adults. But, but then again, we do see some associations for the other cell types as well. So there might be other things, but of course, development during gestation and aging might not be completely overlapping. Anyway, so yeah.

Hannah Went (29:05.399)

Hannah Went (29:16.738)
Sure. Yeah, that was definitely.

Kristine Løkås Haftorn (29:18.296)
But it might be some of the reason for not having this. Yeah, for not seeing kind of the same pattern all over from conception to death, I guess.

Hannah Went (29:30.018)
Sure, yeah, that was my next question was, I know you talk about this in a portion of your paper as well, because I really want listeners to understand the importance of your work, right? And why we’re looking at all of these markers. So, number one, we’re seeing these nucleated red blood cells really associated with DNA methylation and gestational aging. What you then found is this may be the reason we see a poor correlation between epigenetic age clocks and adults versus newborns.

right? Because in those nucleated red blood cells, like you just mentioned, we’re seeing those stay around for what, maybe a couple days after birth and then seeing those go away. They’re becoming mature red blood cells. So that might explain the poor correlation there. But what about then epigenetic age clocks for newborns? Are we going to have to create new clocks then, new epigenetic age clocks?

Kristine Løkås Haftorn (30:00.356)

Hannah Went (30:27.982)
Or do you know maybe when the threshold at a particular age, the clock becomes better? That’s obviously a very open-ended question as it’s going to be different for every single epigenetic age clock. But could you theorize about that?

Kristine Løkås Haftorn (30:39.692)
Yeah, I mean, there has been done some work on children. Not a lot, but some work, epigenetic aging in children and adolescents. And I’m not sure how much of this is published, but I know it has been done some work where you see this gestational age clock. It works a bit.

like maybe the first year or so and then it kind of just stops working at all. While for the age some of the age clocks you get maybe a little bit at least you get like a negative value of some sort when you try to look at the kind of preterms or something like that. And then for children you see that it’s kind of

Hannah Went (31:25.358)

Kristine Løkås Haftorn (31:37.92)
something in between the gestational age clocks and the adult clocks. So it seems to be some kind of kind of development going on and probably, I mean what I think is that it’s different CPGs that are important in different kind of parts of life and maybe after 20 it stabilizes a bit because it’s not that much development going on.

Hannah Went (31:58.68)

Kristine Løkås Haftorn (32:05.244)
After 20, it’s more aging. But I mean, it’s difficult to say, but it’s a lot of interesting work going on right now, trying to disentangle these different clocks and different processes going on.

Hannah Went (32:08.369)

Hannah Went (32:26.75)
Yeah, no, definitely. And Christine, I’ll leave the listeners with this again, just really trying to drive the importance behind this work. And then we’ll switch gears to another subject that I’m equally as interested in. But with a better gestational age clock, if we are able to develop a really great clock, and I think you’ve already done a lot of that work, how would we actually use it to look at developmental timing? And do you think that would eventually

could improve pregnancies as well.

Kristine Løkås Haftorn (32:58.764)
Yeah, I mean, as I mentioned earlier as well, you know, right now we’re only able to predict gestational age at birth. So in countries where ultrasound is available and pregnant women are followed up closely, this may not be so useful clinically right now. There are of course several countries where gestational age of the child is not known at birth at all, where this maybe could be useful to help.

Hannah Went (33:05.934)

Hannah Went (33:23.713)

Kristine Løkås Haftorn (33:27.66)
decide what kind of care and intervention the newborn needs. For instance, if there’s preterm birth, there are some complications. Of course, an interesting point about these nucleated red blood cells is that some of them escape into the mother’s bloodstream during pregnancy. So, and these can be isolated. That’s already done for genetic testing.

Hannah Went (33:53.454)

Kristine Løkås Haftorn (33:53.992)
So one possibility is maybe in the future would be to develop these nucleated red blood cell specific clocks that can kind of track the epigenetic gestational age during pregnancy. But of course, it’s a lot of a lot of things you have to figure out before we’re there. And we also know that some newborns have a larger discrepancy between.

Hannah Went (34:13.812)

Kristine Løkås Haftorn (34:19.64)
gestational age predicted by DNA methylation and gestational age estimated by ultrasound. So one hypothesis is that this tells us something about the developmental maturity of the newborn, kind of like age acceleration, epigenetic age acceleration, and that if we get kind of, if we can make a clock that really gives us a good, that is really a good marker of developmental maturity.

Hannah Went (34:25.495)

Hannah Went (34:34.549)

Kristine Løkås Haftorn (34:48.568)
this might be very useful to, yeah, for example, in preterm births, what kind of interventions or care do they need? Because some babies are maybe born one month early and they have a very preterm phenotype, while others are basically mature.

And what we’re thinking is that maybe the epigenetic kind of measures or the epigenetic clocks can tell us something more than what we can tell just by the gestational age that we can measure from ultrasound or things like that. But I think definitely that it’s a lot of opportunities to explore these clocks more and see what we can actually use them for.

Hannah Went (35:46.278)
I’ll make sure to follow up with you on that. So yeah, I think it’ll be exciting. And I know a lot of people are interested in newborn care, newborn screening. And if we’re able to do that through epigenetic DNA methylation, I think it would only give us some more insights as well. So we’re…

Kristine Løkås Haftorn (35:48.527)

Kristine Løkås Haftorn (36:02.28)
Yeah. And then there are actually also, I have to mention that because there are quite a lot of epidemiological studies now that are looking at this gestational age acceleration and kind of looking at different phenotypes and finding different, yeah, some differences between the children that have like an accelerated gestational age. So there is…

Hannah Went (36:10.862)

Kristine Løkås Haftorn (36:28.356)
Definitely a lot of work going on there as well, trying to find out what we can actually see with this measurement.

Hannah Went (36:34.598)
Using the clocks that you’ve that we’ve been talking about you mean, okay, gotcha Yeah, I’ll have to do a deeper literature dive there and maybe see the associations You know what type of epidemiological factors are associated with? That faster gestational aging which I’m sure we can talk about and assume some of those but I’ll do I’ll do a deeper literature dive there What I want to switch over to now Christine is art

Kristine Løkås Haftorn (36:37.1)
Yeah, exactly.

Kristine Løkås Haftorn (36:52.215)

Hannah Went (36:57.61)
which you mentioned earlier. So you’ve done some work in this, and I think this becomes really interesting to talk about as it relates to epigenetics. What is art? And so just to find that again, I know you did previously, let’s lay that out for the listeners, and then is this the same thing as IVF, or is this different?

Kristine Løkås Haftorn (36:58.656)

Kristine Løkås Haftorn (37:16.568)
Yeah, so ART or ART, it stands for assisted reproductive technology. It’s a collective term for medical procedures where either eggs or embryos are handled outside of the body. And these treatments or procedures are usually for treating infertility. And IVF is one type of assisted reproductive technology. So that’s in vitro fertilization where…

kind of where the egg cells and the sperm cells are combined in the laboratory to kind of, yeah, fertilize in the lab.

Hannah Went (37:59.486)
Yeah, and do we see, again, laying out kind of the groundwork for everyone here, do we see differences in disease outcomes and I don’t know if you want to call them, you know, art babies and traditional birth babies or what the correct nomenclature is there? But do we see differences? What do we know or what does literature tell us about that so far?

Kristine Løkås Haftorn (38:11.192)

Kristine Løkås Haftorn (38:20.488)
Yeah, I mean, ART is associated with a higher risk of some adverse outcomes in the child. And that includes fetal growth restriction, preeclampsia, some birth defects, also some rare imprinting disorders. And I think it’s important to note that these risks are modest. So it’s not like a huge risk to get ART, but we do see some increased risks in some.

some of these children and some of these risks also have been considerably reduced the last years because it has become much more common to transfer only one embryo back to the mother at the time instead of several. So before it was much more common to transfer multiple embryos at the time and then of course the twinning rates were much higher.

And twinning in general is associated with higher risks of a lot of different complications. So that definitely helps to only have transfer one embryo. And also these risks seem to differ between different procedures. So for example, children born after what we call fresh embryo transfer. So this is when the embryo is transferred without being frozen first.

Hannah Went (39:18.978)

Kristine Løkås Haftorn (39:46.968)
It’s just it has developed in the lab and then it’s transferred back to the mother when it’s ready. These children have a higher risk of low birth weight, whereas the children that are born after frozen, frozen thawed embryo transfer. So that’s when the embryo is frozen and then at a later point thawed and then transferred back into the mother.

And those children, they have a higher risk of higher birth weight. So it’s kind of a, yeah, the other way around and we don’t really know why. But that’s one of the things that we think that maybe can be due to DNA methylation differences. So we know a little about health later in life in ART conceived babies, because most of these

Hannah Went (40:24.555)

Hannah Went (40:40.683)

Kristine Løkås Haftorn (40:42.84)
children are still very young. But there are some studies that point to a higher risk of, for example, diabetes and cardiovascular disease. And there has also been some reports on a higher risk of some neurodevelopmental adversities, but there’s conflicting evidence on this. And the problem is generally to small sample size and to short follow-up time.

Hannah Went (41:10.146)
Sure, we’ll need more of that longitudinal data analysis. And I think these assistive reproductive technologies, I think, are very new as well. So I’m sure this will really be a very large area of research. And that was interesting that you mentioned with the fresh, I guess, embryos where they’re not being frozen and thawed. They have the low birth weight, and then the frozen and thawed embryos, they’re having higher birth weight. So with those different weight outcomes as well, I think we can maybe make.

Kristine Løkås Haftorn (41:12.781)
Sure. Yeah.

Hannah Went (41:38.694)
hate to say assumptions, but that each of those low birth weight is going to be associated with a number of different outcomes versus that higher birth weight as well. So we’ll have to follow up with those children and see those definitive disease outcomes. And then again, maybe it’s a matter of more information and choice to the mother, to the family as well to proceed with these technologies. So there are…

Kristine Løkås Haftorn (41:48.737)

Kristine Løkås Haftorn (42:03.232)
Yeah, sure.

Hannah Went (42:05.478)
Yeah, and then you mentioned some of the DNA methylation association. So I know a lot of epigenome-wide association studies of art have been published. What do we know about this prior to the research that you’ve done? Are there any other associations that you want to talk about? You just mentioned a couple of disease ones.

Kristine Løkås Haftorn (42:22.528)
Yeah, I mean, DNA methylation is particularly interesting to study in ART pregnancies because the timing of these procedures, they coincide with the epigenetic remodeling that I mentioned earlier. So where the DNA methylation patterns are removed and then re-established in the early embryo. And so the question is, of course, whether these processes are disturbed in some way.

by the ART procedures, especially since we also see this increased risk of imprinting disorders and imprinting is a DNA methylation process. So as you mentioned, there has been published a number of EWASs the last few years. The first ones had kind of a problem with very low sample size, and then it’s very difficult to kind of say much about. I mean, it’s a…

limited power to say anything about the results. But many of those IWOS’s have reported differences in DNA methylation at birth between ART-conceived and naturally conceived children. And last year, our group published this large epigenome-wide association study where we also had epigenetic information on the parents of the children because it’s been a big

Hannah Went (43:20.341)

Kristine Løkås Haftorn (43:47.408)
question whether these differences that we see between ART and non-ART babies could just be due to differences in subfertility between the parents. So that the differences in fertility and therefore there are some DNA methylation differences between the parents and that’s kind of something that children inherit.

Hannah Went (43:59.959)

Kristine Løkås Haftorn (44:14.6)
So what we did here was that we also, or we had the opportunity to actually control for these DNA methylation patterns of the parents. And we found quite a lot of methylation differences. So more than 600 CPGs had differential methylation between RT and naturally conceived children. And most of these also persisted after controlling for

parental DNA methylation. So it seems to be actually due to the AOT procedures and not due to the parental subfertility.

Hannah Went (44:50.93)
Got it, which is a good differentiation to make here, so we can continue to study and look into it. So yeah, you’ve done these studies, you’ve noticed that there’s a difference. None of the studies I guess I’ve known to date until yours have actually investigated those CPG regions on the X chromosome. So how did you do this in particular? What did you find on the X chromosome?

Kristine Løkås Haftorn (44:53.726)

Kristine Løkås Haftorn (45:13.696)
Yeah, so this was a study that was led by a colleague of mine, Julia Romanowska, and the X and also Y chromosomes are usually excluded from EWASs in general, not only the ART EWASs, but also other phenotypes, because they complicate the analysis quite a lot, because you have to account for the X chromosome inactivation and also the genes that escape this X chromosome inactivation.

And there is a gradual loss of X chromosome with H. So that might complicate analysis where you have different H groups. So in an X chromosome wide association study, it’s important to analyze girls and boys separately because of these very different overall DNA methylation profiles for girls and boys on the X chromosome. Since girls have two X chromosomes, we have the…

X chromosome inactivation while boys only have one X chromosome. So we investigated the CPGs on the X chromosome in girls and boys separately and searched for differentially methylated regions and regions of co-methylation that flanking those significant CPGs. And since we use the same data set as in this larger EWAS on the autosomal chromosomes,

we were also able to adjust for parental methylation here. So we were also very lucky to collaborate with a group in Australia who provided an external cohort where we could replicate our findings as well.

Hannah Went (46:54.558)
Yeah, and so you found, so I guess just summarizing there, what did you actually find on the X chromosome? Are you just looking at those differentiated methylated regions? Did you look into those CPGs and the, I guess, mechanism of action?

Kristine Løkås Haftorn (47:07.156)
Yeah, I mean, we did find some differential methylation in both girls and boys. More differentially methylated CPGs and regions in girls than in boys. And there was actually no overlap in the significant findings between the sexes. So it seems to be a very sex specific association. But these, it wasn’t as clear in this paper where those CPGs kind of

Hannah Went (47:33.666)

Kristine Løkås Haftorn (47:36.664)
belong, what they’re doing, how they kind of fit into the whole picture. Um, what was interesting was that several of these CPGs in regions were located in or near genes that are expressed in tissues that are relevant for ART and sex, so like testis, placenta, fallopian tube. Um, and, um, the other thing we saw was that several of these genes were also associated with neurodevelopment and an intellectual disability.

Hannah Went (48:04.931)

Kristine Løkås Haftorn (48:06.88)
So as I mentioned, there has been some studies that have found an association between those outcomes and ART, but they’re not really very clear associations. So, but anyway, it’s I think it’s an interesting kind of thing to follow up on to see if we can understand more about this.

Hannah Went (48:29.334)
Yeah, and the differences. Yeah, I actually didn’t know, I guess I may have assumed, but I’ve never really talked about it, that the X and Y chromosomes are not involved in EWAS studies. I mean, that would make sense, right? I’m sure there are a lot of complications when you’re including those. So you said that the X and Y chromosomes were not included because of X activation and that there are gonna be age differences even when you’re looking at that X chromosome and you’re losing part of that X chromosome over time. Is that correct?

Kristine Løkås Haftorn (48:56.684)
Yeah, exactly. Yeah, so it makes sense to kind of do the autosomal chromosomes in one analysis and then do a separate analysis on the X chromosome and then do that separately for girls and boys.

Hannah Went (48:58.774)

Hannah Went (49:02.112)

Hannah Went (49:08.778)
Right, right, okay, perfect. And then maybe too early to tell, I think, but I still think this is really important research. How might this affect ART protocols in the future? What type of information can people use from this type of analysis for assistive reproductive technology if they’re choosing to go with that path?

Kristine Løkås Haftorn (49:28.832)
Yeah, I mean, so far we don’t really know what part of the procedures affect DNA methylation and we don’t really know how severe consequences this has as well. So, I mean, ART may include so many different procedures like…

Hannah Went (49:44.803)

Kristine Løkås Haftorn (49:52.484)
hormonal stimulation of the ovaries. We have like the surgical retrieval of the oocytes. And then we have the fertilization step, which can be with or without directly injecting a sperm into the oocyte. We have culturing, storing, transferring the embryos and the cells involved are exposed to changes in temperature light.

Hannah Went (50:09.774)

Hannah Went (50:16.526)

Kristine Løkås Haftorn (50:19.628)
oxygen levels, pH, different types of culture media, and many different things. So I think there’s a lot more studies needed to kind of pinpoint what part of the procedures that actually may have an impact on DNA methylation, if any, and also more studies on long-term health effects. And when we have a clearer picture of that, it might have some clinical consequences. But I think it’s…

Hannah Went (50:36.194)

Kristine Løkås Haftorn (50:45.856)
important to kind of mention that it’s… Yeah, so like I said, the risks for disease that we see so far, they’re not very big and like it seems to go okay with these children. I mean they seem to be fine so I think it’s important to not kind of… Of course we should be careful and of course we should study this and figure out if something should be done differently but it’s also important not to

Hannah Went (50:57.558)
Mm hmm.

Hannah Went (51:02.125)

Kristine Løkås Haftorn (51:16.598)
about this because I don’t think it’s reason to be very scared about this.

Hannah Went (51:21.194)
Yeah, I’m glad you said that as well. Yeah, I think the main reason for this research and the ART is, hey, how can we maybe make it a little bit better, right? And make those outcomes just a little bit better, maybe keep people from even worrying as well. So I know there are a ton of variables and different factors like you were just talking about.

Kristine Løkås Haftorn (51:32.099)

Hannah Went (51:41.218)
you know, temperatures, different transfers, different parts of the laboratory procedure. So even if we can get just a little bit better and understand what each part of those processes means in terms of outcomes, I think that can even give people a little bit more relief or, you know, hope as well. Um.

Kristine Løkås Haftorn (51:55.36)
Yeah, definitely. For sure.

Hannah Went (52:00.538)
So thanks, Christine. We’re getting toward the end here. This has been amazing. We’ve been talking about all things epigenetics, DNA methylation, gestational age clocks, the assistive reproductive technology, maybe what we’re able to understand between kind of what these DNA methylation markers are telling us with all of these different outcomes. I kind of have a curve ball question. I always ask my…

Kristine Løkås Haftorn (52:03.27)

Hannah Went (52:24.304)
uh, my people on, on my podcast at the end, uh, if you could be any animal in the world, Christine, what would you be and why?

Kristine Løkås Haftorn (52:30.948)
Yeah, that’s funny because I’ve actually been asked that question before in my friend’s wedding. They sent out this question before their wedding to all of their guests and then we answered that and they actually made for their wedding, they made this small figure of the animal that you said you would be and they placed that on kind of where you were supposed to sit during dinner. That was a really cute thing.

Hannah Went (52:35.171)
Oh cool!

Hannah Went (52:42.769)

Kristine Løkås Haftorn (52:59.032)
So my animal is Pangolin, which is this very cute little, I guess, ant eater or something from South America. They’re super cute. They have these scales and when they’re scared they curl into this tiny ball. And they also look very funny because they are actually walking on their hind feet. And yeah, so really cool.

Hannah Went (52:59.068)

Hannah Went (53:04.412)

Hannah Went (53:24.762)
Yeah. How do you spell it?

Kristine Løkås Haftorn (53:28.164)
P-A-N-G-O-L-I-N, I think. Hahaha, yeah.

Hannah Went (53:35.03)
Okay, I was close. I’m gonna look it up after this because I’ve never heard of it. But that’s, no, that’s so cool. And that’s a good story too. I like with the wedding and how they created those little animals. So no, that’s great, Christine. Well, I really appreciate your time. I’m gonna link, of course, all of the papers and even links to your profiles or whatnot. But for listeners who want to contact you, where can they find you?

Kristine Løkås Haftorn (53:59.744)
Yeah, I guess the easiest thing is probably by email So I guess you’re probably you can put that out as well Maybe and or yeah Twitter or something like that. I’m on Facebook as well. So yeah

Hannah Went (54:04.099)

Hannah Went (54:09.759)
Yeah, definitely.

Hannah Went (54:13.91)
Perfect, yeah. Awesome, Christine, I’ll link your Twitter as well. Well, we’ve come to the end of this amazing podcast interview again. Thank you so much, Christine, and thank you everyone for joining us and listening at Everything Epigenetics podcast. Remember, you have control over your epigenetics, so tune in next time to learn more on how. Thanks again, Christine.

Kristine Løkås Haftorn (54:33.412)
Thank you.

About this Guest Expert

Kristine Løkås Haftorn
Kristine Løkås Haftorn is an expert in gene expression regulation in chronic fatigue syndrome/myalgic encephalomyelitis, and is currently focused on studying the relationship between gestational age and DNA methylation in cord blood for her Ph.D. at the Norwegian Institute of Public Health’s Centre for Fertility and Health.

More About me

Everything epigenetic
Everything epigenetic
Epigenetic Gestational Age Prediction

More Episodes