Cheek swabs, epigenetic clocks, and kids

Cheek swabs are easy to collect, but the science behind them is anything but simple. In this episode, Dr. Sarah Merrill and Dr. Chaini Konwar discuss how the proportion of cells collected from cheek swabs — buccal epithelial cells — change with age and why this matters for pediatric epigenetics. 

Learn how these changes impact biomarker accuracy, epigenetic clocks like PedBE, and our understanding of child development and health risks. For scientists and curious minds alike, the episode explores how cheek swabs — from infancy through adolescence — can help us better track, understand, and support children's health over time. 

Learn more about topics discussed in this episode:

Impact of age-related changes in buccal epithelial cells on pediatric epigenetic biomarker research, Nature Communications

All episodes written and produced by the Research Communications team at BC Children's Hospital Research Institute.

Theme music: "Life Is Beautiful" by Anastasia Kir

Transcript

Kristen Hovet (00:00)
My name is Kristen Hovet and I'm a research communication specialist for BCCHR. Joining me today are Dr. Sarah Merrill and Dr. Chaini Konwar. Dr. Merrill and Dr. Konwar co-first-authored a research article titled, “Impact of age-related changes in buccal epithelial cells on pediatric biomarker research,” published in January 2025 in the journal Nature Communications.

Sarah Merrill (00:28)
I'm Sarah Merrill. I was a postdoctoral research fellow in Dr. Michael Kobor's lab at BCCHR, and I am now an assistant professor and a PI of the Epigenetics and Psychosocial Interventions Lab at the University of Massachusetts Lowell.

Chaini Konwar (00:47)
Hi, I'm Chaini Konwar. I'm a bioinformatician in Dr. Michael Kobor's lab at BC Children's Hospital Research. I have been working for over five years now alongside a fabulous team of data scientists who routinely works with high-dimensional epigenomic and genomic data. I have a PhD in medical genetics from UBC where I was working on placental epigenetics and genetics with Dr. Wendy Robinson at BCCHR.

Kristen Hovet (01:20)
For listeners who might not be familiar, why do cheek swabs matter in children's health research and what can buccal epithelial cells or BECs tell us?

Sarah Merrill (01:33)
Great question. These cheek swabs are really useful for kids because they're not painful. They're not scary. They're not uncomfortable even really; it's just a little plastic brush up against your cheek. We can use it in kids as young as infants. And we also have the benefit of cheek swabs not running into the same kind of cultural and religious taboos that we find with both saliva and blood collections, which are other tissues that are commonly used for this type of research.

In terms of the actual collection of the sample, it’s really useful specifically for a children's population and looking at children's health. And because of this, we have a large body of work looking at how we can predict things and these samples can reflect things that have happened to children to improve their health. So, things like exposure to stress or pollution or chemicals. That can also be useful for screening things like cancer or heart problems.

We study DNA methylation, which is a type of epigenetics, where we can look at these tags that are on the DNA. And they don't affect the genes themselves, but they can affect how the genes behave. And they can also act as biomarkers of things that have happened in the past biologically. We can use these to look at children's health in terms of their actual experiences, as well as their likelihood of having a health problem in the future or even if they are currently having a health problem at that moment.

Kristen Hovet (03:22)
Your study looked at cheek swabs from 12 pediatric cohorts, comprising more than 4,600 children from babies to teenagers. Were there any surprises as the data started to come together for such a large and diverse group?

Chaini Konwar (03:39)
Yes, thank you for this question. Well, first, I would definitely like to thank our participants and our wonderful collaborators who shared their cohorts with us. It really allowed us to create this comprehensive database of cheek swabs from approximately almost 5,000 typically developing individuals, spanning across the entire pediatric range, which was from two months to 21 years old.

And some of the cohorts were even actually longitudinal, meaning we had measurements from multiple time points that allowed us to look at this data in a way. Now getting back to what was one of the most surprising or exciting bit that we found was initially at the start when me and Sarah started to conceptualize our work in some of the preliminary analysis, what we hypothesized — probably in a naive way — is that the linear decrease or like the decrease of epithelial cell proportions with age, we just expected it to be in a linear fashion across all our participants, regardless of where they fall in the developmental window.

However, one of the wonderful reviewers in Nature Communications pointed out that there is a subset of samples which did not seem to follow that trend. And while addressing the reviewer comments, when we inspected our samples, we found that a subset of samples around adolescence actually did not conform to the trend. And then we started working with some other sophisticated statistical approaches, which allowed us to examine samples from different developmental periods a little differently.

That's when we defined a transition from childhood to using a threshold of 10 years. Again, as stated in the paper, this transition is not about a single age per se, but more around a developmental period, which will encompass the ages at 10 as well. So we did our analysis for the nine-year-old cutoff and also 11-year-old cutoff. And what we saw when we looked at the subset of samples from infancy to childhood, less than 10 years old, the decrease in epithelial cell proportions from the cheek swabs with age was more pronounced and stronger and was in a very linear way. 

However, we did not see that in the adolescents, which is around the pre-adolescence and the end of adolescence age. And also when we looked at variability, children in the adolescent [group] showed much more variability in buccal epithelial proportions compared to the infancy to childhood [group]. So all of these were pointing towards [the fact] that if you're looking at cell-specific associations with age, it is very important to consider the developmental period. And since then, we really acknowledged the relevance of the developmental periods and subsequently always performed our analysis appreciating that sort of distinction.

Kristen Hovet (06:51)
Thank you. And I think this question will probably overlap with what you just responded with. But one of your key findings is that the number of buccal epithelial cells drops as children age then level off during the teen years. Why is this important? And what could it mean for interpreting epigenetic data in pediatric populations?

Chaini Konwar (07:12)
Yeah, we do have some theories and speculation as to why we're seeing the differences depending on developmental periods. Some rodent data says that the morphology — the size and shape of the oral epithelial cells — changes over time. There is a reduced proliferation. The size may increase, but there is reduced density. Another compelling [finding] is the presence of pediatric gingivitis, which is an inflammation of the gums reflecting bacterial challenges. And this condition is almost absent or the presence of such bacterial species in the gingival tissue is almost absent in children, but are present in the mouth of adolescents. Conditions like gingivitis or some other oral health events, oral hygiene, diet, use of orthodontic appliances, apparatus, these all can contribute to the proportions that we have observed.

Sarah Merrill (08:09)
And we also see in the quite young part of our childhood, sort of zero to 10 part of our samples that have that significant and reliable decrease in buccal proportions, that this could also be related to normative craniofacial development, that as their faces and their heads get bigger, as they are wont to do, that this also affects the proportion of skin to immune cells, essentially in the samples that we're seeing in a reliable way that you don't necessarily see in the adolescents where you aren't seeing the same kind of facial development, but you are seeing a lot of variability from things like Chaini said, you know, oral health, gingivitis, things like that.

Kristen Hovet (09:00)
Things like baby teeth, braces, as you mentioned, even how often kids visit the dentist might affect the cheek swab results. What do you want other researchers to know about this and even parents? What can they take away from this?

Sarah Merrill (09:15)
Yeah, I think that's a great question. And for example, in our study, in one of the cohorts that had contributed data, we had information on teeth brushing and dental visits. And what we found is we didn't see a significant difference in terms of brushing their teeth more often. These are very young children. We were looking at two- and three-year-olds.

So this is not necessarily true for older children, but at least in young children, we didn't see a huge difference in the amount of teeth brushing, but generally in our population, they had good oral hygiene. But we did see a difference in people who did have regular dental visits in early life versus those who didn't. And the people who didn't have those regular dental visits were more likely to have more cells, essentially, in their mouth in comparison to the skin cells that we — much in line with what some of the other research has found, for example, with braces — that because there are more opportunities for bacteria to hide in the mouth, more things going on in the mouth, realistically it’s harder to get into those spaces with brushing and things like that, that you see a similar difference with more immune cells in the mouth and less of the buccal skin cells. 

And so I think what we would like researchers who are also using these cheek swabs to know is that it's important to ask questions like this. Things like oral hygiene, dental visits, you know, especially if you're working in populations where orthodontia is common, not just braces, but permanent retainers, nighttime retainers, things that people wouldn't normally think to report to you as a potential health difference. These are things that you could then be able to control for in your models or account for, especially if you are concerned that it might be associated with variables that you think are important for your variables of interest.

For example, a lot of epigenetic work is interested in the overall effects of socioeconomic status, which can encompass lots of things from income to education to your social position, right? And these can overlap with access to something like dental visits or orthodontics, which can be very expensive. And you wouldn't want to accidentally essentially confound these differences in cell types because of the types of immune cells and bacteria in the mouth with something that is actually an income-related issue.

Kristen Hovet (11:52)
You worked with something called an epigenetic clock, which estimates biological age as separate or different from a person's actual chronological age. How do these clocks work and how does your study help make them more accurate for studying kids' health?

Sarah Merrill (12:09)
Great question. Epigenetic clocks are very popular these days. So an epigenetic clock measures your biological age or the age of your cells, essentially. And they use those patterns of tags, these chemical tags on your DNA that don't change your sequence but can change the function of these cells. And across many sites where these tags can happen, we can look at them and we know that they change reliably across people with age. By measuring how much of that tag you have on your cells at any particular point in time and adding those together, we can estimate approximately how old you are biologically based on these epigenetic markers. Then we can compare that to your chronological age, your actual age based on your birthday.

And biological age can be higher or lower than your actual age, right? And that can reflect differences in your health, how your body is aging, how your body is developing outside of just your calendar number of years. And it is important to say that a lot of the work that has been done on epigenetic age is primarily in adults, right? Where they have found that having an older biological age in comparison to your actual age is a clear health risk. It puts you at a higher risk for many conditions, cardiovascular disease, diabetes, even an earlier death potentially. And a younger epigenetic age is always a good thing, basically, right? That being younger is good. But in kids, kids don't just age, they develop, they mature.

But there is increasing evidence in children that being either substantially biologically older or younger, similar to both precocious and delayed development, is not necessarily healthy for kids and that what you're really looking for is to be as old as you are — that your biological age is about as old as your actual age. Things like an experience of adversity, experiences of abuse, for example, high levels of stress, etc. have pretty consistently been associated with having an older epigenetic age in children, which is also associated with health risks. So being able to measure this as a marker in children accurately can be very helpful, both in helping us understand how much of a difference is meaningful and in what direction, which is a question we are still answering, and what does that really convey for risk.

And in order to do that, we need the most accurate measures possible. So by accounting for these differences in the number of buccal cells to immune cells in your cheek swabs, as we recommend in this paper, we are able to have more and more consistent measurements from these samples in kids, which will help us to be able to interpret and use and even develop new biomarkers better.

Chaini Konwar (15:35)
And Sarah has wonderfully explained the epigenetic clocks for us and their relevance. What I would like to add is maybe think of epigenetic clocks as measuring the wear and tear on a pair of shoes. Imagine that as you walk in your shoes, they slowly start to show these signs of use, you know, the creases and the wear on the soles. The more you walk, the more signs of aging on the shoes will show. So overall, what's important is the number of steps that have been taken wearing those shoes, not the calendar date of when you bought them. 

Similarly, as you age, the stress and adversities, one's experience make the cells older than our calendar age, which then increases our risk of various diseases. But this is, as Sarah mentioned already, relatively straightforward in the adult space. But as we have mentioned a few times now, the kids are not aging, they are developing. And so we're slowly starting to understand what this means. What do these clocks mean, these epigenetic clocks in the pediatric space?

Kristen Hovet (16:43)
You used one that the Kobor Lab developed, right? Am I saying it right? Is it Ped-bee or Ped-B-E? 

Chaini Konwar (16:51)
Yes. That's the start of our paper, our PedBE, which is a pediatric buccal epigenetic clock, which is sort of, yeah, the abbreviated version. And that is supposed to provide the most accurate estimation of calendar age or chronological age in pediatric cheek swab samples. And there are a multitude of clocks which are developed for adult aging populations, but clocks in the pediatric space are really limited and PedBE definitely is one of those.

Kristen Hovet (17:23)
I'm going to ask another question that kind of talks about that transition around age 10. I know you talked about it already, but it's really an interesting part of the paper for sure. So your study highlights that developmental transition that happens around age 10. How might this transition in the context of buccal epithelial cell proportion influence how researchers design the clocks themselves and epigenetic studies.

Chaini Konwar (17:51)
I think I can start a little bit on this and then Sarah can chime in for sure. But what we are really saying, using that cutoff, is yes, we used a cutoff of 10 years to do most of our subsequent analysis, but really what we are emphasizing is that transition from childhood to adolescence. So it is not really about a single age per se, but it's that entire developmental period, which is why we tested most of our analysis using a cutoff at nine or a cutoff at 11. But to sort of acknowledge that what is happening from infancy to childhood could be really different, as we observed, and distinct from what's happening in adolescence. Like we see this decrease in buccal epithelial cell proportions with age very pronounced, very striking in a linear fashion from infancy to childhood. But then when we only narrow it down or restrict our samples to those around adolescence, which is like pre-adolescence and end of adolescence from like 10 to 20 years, we don't really see that association. So just acknowledging that, you know, when you're examining these cell-type associations with age, acknowledging the developmental period and taking that into consideration is what we are suggesting.

Sarah Merrill (19:03)
We also saw that, in the children who had a strong negative association with age and buccal proportion, with our youngest participants having the highest number of buccals in their cheek swabs, this was a strong relationship, where your essentially cell-type proportions in your sample and therefore the methylation that you're able to measure, right, because it is entirely cell type dependent, is completely confounded with age. So if you would like to do a longitudinal study over the course of early childhood — very popular, a great question from a developmental origins of health and disease perspective — you definitely want to make sure that you are accounting for these cell-type proportions, even when using, you know, cumulative biomarkers that were designed in these samples, because by virtue of looking over time, you have an inherent confounding with the cell-type proportions that you'll have. 

And then alternatively in adolescence, what we actually found was a lot of inter-individual variability. So there wasn't a strong relationship or any kind of confounding between cell-type proportions and age. However, when we corrected for the differences in cell-type proportions, we actually found stronger signals, where those cell-type proportions — because they were so variable in the adolescents — accounting for them essentially eliminated noise in our analyses that made our associations stronger. And so actually, accounting for these buccal proportions can be important even when you're not worried about it being a confound, because it may be actually responsible for some of the variability that you're not particularly interested in in your variable of interest. And so modeling this, even though the relationship is quite different between children and adolescents, modeling with epigenetically estimated cell-type proportions in mind — no matter where along the pediatric health spectrum you are studying — can be helpful for clarifying what the actual relationships between your variables are.

Kristen Hovet (21:17)
And I think I recall seeing an email from Dr. Stewart, who was talking about OCD research. So is there a paper on epigenetics and OCD coming out soon?

Chaini Konwar (21:28)
Yes, stay tuned. So yeah, we had the opportunity to collaborate first with Dr. Evelyn Stewart from BCCHR, where we looked at these OCD kids and looked at how PedBE was able to capture a signal of that. And what we observed was a really cool observation, but also which we have already shown with children affected with autism as well.

So what we saw was that cheek swabs, which are obtained from OCD-diagnosed children, showed increased acceleration or were more biologically mature compared to non-OCD healthy controls. But what was most interesting was this magnitude of association was attenuated when we controlled for the proportion of buccal epithelial cells. Still significant, like the association was still there. It's just the strength or the magnitude was a little different.

And the paper that's coming up is not looking at the clocks, but another facet of epigenetic research, which is around epigenome-wide association studies. And I don't want to give any spoilers, but some cool stuff is coming up soon.

Kristen Hovet (22:38)
What excites you most about where your research is headed next?

Sarah Merrill (22:44)
I am particularly interested, and this is probably not a surprise given the name of my brand new lab, in what this can mean for interventions, especially in kids. I have shown in some of my recent work also with Dr. Kobor and Chaini and the team, in collaboration with some of our other great researchers on our team, such as Dr. Nicki Bush at UCSF and Dr. Justin Parent at URI, that looking at oral samples, both before and after interventions, and sometimes even much later — a year after these interventions have taken place — we can see differences when accounting, and especially when accounting for these buccal cell proportions the way that we recommend in this paper, that these interventions seem to be really genuinely making a biological difference these children. 

For example, with Nicki Bush at UCSF, we looked at children with high levels of trauma who had experienced pretty significant adversity. Some had child-parent psychotherapy, which was an attachment-based therapeutic intervention that is designed to improve the relationships between caregivers and their children after traumatic events have taken place and increase the feelings of trust and safety again. And what we found was those who had had the benefit of child-parent psychotherapy versus matched community controls who had experienced trauma but did not have the benefit of psychotherapy, they actually did not continue to have this biological age acceleration where we saw that after the therapy, there was no increase in how much faster they were aging. 

However, the people who didn't receive the intervention were aging much faster than they were at baseline, and especially a year later, much faster. And that is also something that we have seen recently in a study where we looked at Jordanian adolescents with another collaborator, Dr. Rana Dajani, that we just recently presented at the Human Biology Association, where immediately after the intervention, we saw that those who didn't receive the intervention among these high-risk Jordanian adolescents in Jordan, they were aging much faster and those who received a resilience-building, skill-building intervention were not. They were aging sort of the same as they were at baseline. 

So being able to use methods like this and know that the effects that we're seeing — especially longitudinally over time in pediatric populations — are not an artifact of, you know, confounding with age or confounding with cheek swabs and these other facets and factors, makes us a lot more confident to be able to say, yes, we are seeing a biological effect here to the best of our understanding and being able to implement these in biomarkers, at least for me in intervention contexts especially.

Chaini Konwar (25:55)
Yeah, and I'd just like to add that, as a data scientist with a passion for pediatric epigenetic research, these are really exciting times to be in. And what I would really like to see is the application of PedBE and similar clocks in other developmentally relevant childhood disorders. Again, really coming back to the concept that children are developing and not aging, so in certain scenarios in kids being biologically more mature, may not necessarily be a negative thing. And yes, we saw that OCD-diagnosed children and in autism, they seem to be more biologically mature, but we have just started at the tip of this iceberg and we need to know more. So I'm just excited how these tools will be applied across various pediatric conditions and health outcomes.

Kristen Hovet (26:46)
Very exciting, thank you both. And a question that we're asking all of the podcast guests, how does your research help children live their best lives?

Sarah Merrill (26:56)
That's a great question. I think the honest answer is directly, it doesn't. But, what we are doing and what work like this does — characterizations of these samples, how they act across populations, across developmental windows, and really fundamental aspects of the biology of how we use samples like this, how we look at changes over time — are important for building the foundation of what is relatively a young field in the scientific process, especially looking at it and how I usually look at it, which is really looking at social experiences and psychological experiences, and how these can affect us at a biological level, especially in early life to affect health, both in childhood and lifelong.

And by building this foundation, I always joke — all the graduate students in the lab know that I say — we're really putting in the building blocks that maybe no one will see many, many decades from now, but that will be the foundation on which things like individualized and personalized medicine, effective biomarkers that work across populations to predict health and predict treatments and outcomes will eventually be built on.

Chaini Konwar (28:23)
And now we have this comprehensive database of like 5,000 typically developing children, which will allow us to answer more of these questions and probably get at how we can make a more translational difference to children and their health overall.