Childhood adversity and mental health: Identifying opportunities to reduce risk and promote resilience across the life course

November 30, 2022

Information

YCSC Grand Rounds November 29, 2022

Erin Cathleen Dunn ScD, MPH
Associate Professor of Psychiatry
Harvard Medical School

ID9199

To CiteDCA Citation Guide

00:00Good afternoon, everyone.
00:01I think we'll make a start.
00:05So it's great to
00:06see you all here for grand rounds.
00:08Welcome to grand rounds for those of
00:09you who are celebrating last week.
00:11I hope you enjoyed your Thanksgiving and I
00:13hope that everyone had a restful and relaxing
00:16few days towards the end of last week.
00:18And just a reminder about next week,
00:21we'll have compassionate care rounds
00:23here in the Cohen and live on zoom.
00:25As a reminder,
00:26that session won't be recorded.
00:28So please do join us either in person
00:31or live on zoom, or we'll hear from
00:33an expert panel on the complex.
00:36Care needs of patients dealing with
00:38suicidality and disordered eating.
00:40So please do join us for that.
00:43Now in the spirit of the
00:44holiday that we just marked,
00:46I am very thankful to welcome to
00:48have our speaker join us here today,
00:50Doctor Aaron Dunn.
00:51And so Doctor Dunn is an associate
00:54professor of psychiatry and
00:55Pediatrics and Harvard Medical
00:57School and also an assistant
00:59investigator in Mass General Hospital.
01:01And I think it's fair to say that Doctor
01:03Dunn has pioneered the application of life.
01:06Of course,
01:07epidemiological methods to study the
01:09biological embedding of adversity
01:11and the impact of adversity on
01:14adult mental health outcomes.
01:16And now Doctor Dunn has received
01:18substantial support from the
01:20National Institutes of Health,
01:21including the National Institute
01:22of Mental Health,
01:23and has published prolifically,
01:25as you'll have seen.
01:26And just as recently as two weeks ago,
01:28I think you marked your 100th
01:30publication and a nice systematic
01:31review of the of sensitive periods,
01:34the evidence for sensitive
01:35periods of exposure to child.
01:36Our treatment and the prediction
01:37of adult health outcomes.
01:38So hopefully we'll hear a little
01:40bit about that today and a new
01:42area in the Dunlap looking at
01:44teeth as a potential biomarker of
01:46exposure to early adversity.
01:48Again,
01:48very excited to hear more about that today.
01:51So please give a warm child study
01:53center welcome to Doctor Dunn.
01:59I'm impressed, Karen. You did
02:02that all memorized. It's amazing.
02:06All right, so let me go
02:07ahead and share my screen.
02:18OK. I think we're, I think we're good.
02:22Well, thank you everyone for the
02:23opportunity to be here today.
02:24I'm really excited to share
02:26with you more about my work.
02:27As Kieran said, around childhood
02:30adversity and mental health,
02:31I'm going to tell you a little bit
02:33more about opportunities I think there
02:35are to identify risk and promote
02:37resilience across the lifespan.
02:39Can everyone hear me?
02:40OK, OK, perfect.
02:41So I have no disclosures.
02:45So just to Orient us a little bit,
02:48I want to say a little bit about
02:50childhood adversity.
02:51Childhood adversity,
02:51I think is so critical to study because it's
02:55one of the most impactful social determinants
02:57of mental health as well as physical health.
03:00When I think about childhood adversity,
03:02I think about a range of
03:04different kinds of experiences.
03:05These could be events that happen within
03:07the household or outside of the household.
03:10They could be perpetrated by
03:12loved ones or by strangers.
03:15They could be.
03:15Friends that are acute,
03:16they could be chronic.
03:18Some might meet the definition of a trauma,
03:21others might not.
03:22What we know from large scale
03:24epidemiological studies that
03:26have been done primarily in the
03:28United States is that we know
03:30that adversities are common,
03:31so we know that more than half
03:33of all kids growing up in the US
03:35will experience at least one type
03:37of adversity in their lifespan.
03:39We also know that there are large
03:41racial ethnic minority differences
03:43such that kids who grow up from.
03:46Racial and non white families are
03:49disproportionately affected by adversity,
03:51and similarly,
03:52we also know that girls are
03:54more likely to experience some
03:56adversities compared to boys.
03:58The boys are also more likely
03:59to experience some types of
04:01interpersonal violence in particular.
04:03Now I think this following
04:06statistic is both sobering.
04:08These two sets of statistics are
04:11both sobering but also optimistic.
04:13The first is that we know that
04:15adversity is estimated to at least
04:17double the risk of a mental disorder.
04:19Throughout the lifespan.
04:20Now,
04:21it might not surprise you to hear
04:23that childhood adversity is associated
04:25with child onset or adolescent
04:27onset psychiatric disorders.
04:29But we also know that these
04:31adversities are associated with
04:33increased risk of disorders that
04:34onset for the first time in adulthood.
04:37And we also know,
04:38particularly from recent large
04:40scale meta analysis,
04:42that if these effects of adversity
04:44are causal,
04:44they'd explain about 30 to 40%
04:47of the total variability in risk
04:50for mental health problems.
04:51So to me,
04:52I hear that both optimistically sobering,
04:55right?
04:55It's a it's a scary statistic,
04:57but I think it also suggests
04:59opportunities for where our work
05:01can really have potential impact.
05:03A lot of the work that we do in
05:05my group is focused on depression,
05:07which for those of you who may
05:08not be familiar,
05:09is a major public health problem.
05:11So depression is a disorder that's
05:13common throughout the lifespan.
05:14About one out of every five people
05:16will experience an episode of
05:18depression at some point in their lives.
05:20We also know that depression is a disorder
05:23that disproportionately effects women.
05:25So during childhood, boys and girls
05:27experience similar levels of depression.
05:29But something happens in adolescence
05:31where Girl Scout start to outnumber boys.
05:34By a ratio of two to one, and that
05:36disparity persists throughout the lifespan.
05:39We also know that depression is
05:41associated with a host of negative
05:43consequences in the short and long term.
05:46We know it's recurrent,
05:47we know their side effects from medication.
05:49We know that it affects
05:51people's ability to go to work,
05:53to complete school, and so forth.
05:55And I think in its most severe
05:57form is suicide and self harm.
05:59So because depression is so common,
06:01and because it disproportionately
06:03affects large segments of the population.
06:05And is associated with so many
06:08negative consequences.
06:09Hopefully,
06:09it might not be a surprise to learn
06:12that depression is currently the second
06:14leading cause of disability worldwide.
06:16So my group is really focused on trying to
06:19identify ways that we can prevent depression.
06:21And I think that's really critical
06:23because it's a disorder that
06:25strikes when people are young.
06:26And once it emerges,
06:28it tends to be highly recurrent.
06:30So we know that between 20 to 40% of
06:33people who experience depression will
06:36have had their first onset by age 21.
06:39And we also know that about 3
06:40out of every four people with
06:42depression will experience at least
06:44one relapse in their lifespan.
06:46So when I hear data like this,
06:48to me the the the message is we need to
06:51better understand what causes depression,
06:54what's its etiology,
06:55how does it come about,
06:56so that we can use those insights to
06:59then identify targets to identify kids
07:01who might be at risk and prevent the
07:03onset of depression and do that as early
07:06on as we possibly can in the lifespan.
07:08So.
07:09Committed to that goal.
07:10The current focus of my research group
07:13has been organized in these 44 domains.
07:15So we do work on The Who,
07:17how and the when question
07:19around depression prevention.
07:21So with respect to The Who,
07:23a lot of what we do is focused on trying
07:25to identify people at highest risk using
07:27genetic and other markers of vulnerability.
07:30So that's work that we do in in relationship
07:32to the Psychiatric Genomics Consortium,
07:34for example.
07:35We also do a lot of work,
07:37and I'm going to tell you a
07:38lot about this work today.
07:39Around the biological embedding of
07:41adversity and the mechanisms that
07:44might explain how it is that these
07:46stressors and traumas might get
07:48under our skin to shape our health,
07:50the third area is really focused
07:52on sensitive periods.
07:53Try to understand,
07:54are there ages in the course
07:56of the lifespan when our life
07:58experience matters more?
07:59And could that differentially
08:01predict risk for depression?
08:02And then finally,
08:03I'm someone that really believes
08:05in and committed to translation,
08:07so I don't want to just do ivory tower.
08:10Client,
08:10so to speak,
08:10I want to figure out how
08:12to get our findings out to make a difference.
08:14So that's where we've also been doing
08:16work to try to build novel infrastructure
08:19for scientific knowledge translation.
08:20And I'm proud to partner with Josh Rothman,
08:23a colleague in psychiatry,
08:24around a birth cohort work that
08:26we're doing where we're deliberately
08:28from the beginning trying to design
08:30it to not just observe but also
08:32intervene in those participants.
08:33So what I want to do in this talk is tell you
08:38more about two specific aspects of my labs.
08:40Work related to childhood adversity.
08:43So the first is work that we've
08:44been doing to try to identify these
08:46sensitive periods in development.
08:48And then I'll also transition into
08:50telling you more about what we're
08:52doing to try to overcome measurement
08:54challenges that exist in capturing
08:56childhood diversity exposure.
08:58And then at the end,
08:59I'm not a clinician,
09:00but I'll try to talk a little
09:02bit about some of the clinical
09:04implications and applications I
09:06think might exist for this work.
09:08So let me just also clarify at
09:10the beginning because one of the
09:11questions you might have is,
09:13you know, childhood adversity,
09:14trauma, aces,
09:15do all of these things mean the same thing?
09:18So from my perspective,
09:19I tend to use the language of
09:21childhood adversity because I
09:23think it's more encompassing.
09:25Childhood adversity is generally
09:27defined as circumstances or events
09:29that threaten children's physical
09:31and psychological well-being and
09:33their deviations from what you would
09:35expect kids who are typically.
09:37Developing should experience.
09:38I think of aces as being a a set
09:42of 10 markers that have been most
09:44well studied in the context of of
09:47adverse childhood experiences studies,
09:49and so those are sometimes overlapping
09:52with the adversities that we study,
09:54but tend to sometimes not include
09:56all of them.
09:57Some of the adversities we study
09:59could be stressors,
10:00some could be traumas and toxic.
10:02Stress to me really differentiates
10:04the context surrounding the stressors
10:06and traumas.
10:07The kids are going through and
10:09whether or not they have buffers,
10:10mainly those protective adults
10:12who can help buffer those effects
10:14of those stressors for them.
10:16So hopefully that's clarifying in
10:18terms of just getting a better feel
10:20for how I think about adversity.
10:22So in terms of talking about identifying
10:25sensitive periods in development.
10:26So one of the big questions that
10:29I think exists for the field is,
10:31you know,
10:31how does the timing of adversity
10:33shape risk for depression or any
10:36other adverse health outcome?
10:37And if you turn to the literature,
10:39you'll see that there's been a lot
10:40of different theories that have
10:42been proposed on this topic.
10:43So a basic model is an exposure model.
10:46And this model simply states that
10:48people who've been exposed to
10:50adversity have an increased risk of
10:52an adverse health outcome relative
10:54to people who are not exposed.
10:56There's also accumulation models,
10:58and in its simplest presentation,
11:00I'm showing here a basic dose
11:03response relationship.
11:04So the more adversity,
11:05the more at risk you become.
11:07Now that could be exposure
11:09to the same type repeatedly,
11:11or it could be different types of exposure.
11:14There's also recency models and
11:17recency models. Oops, sorry.
11:19Recency models focus on the time
11:21since the onset of the event.
11:26So a recency model says that your risk
11:30of an adverse health outcome is greatest
11:33shortly after you've been exposed,
11:35but then your risk decreases over time.
11:38And then finally there's
11:39a sensitive period model.
11:41And a sensitive period model is
11:43really asking, are there specific
11:44age stages in the course of the
11:46lifespan when our experience matters?
11:48More so over the years on and very
11:52symbolically marking my 100th publication.
11:55I love the symbolism.
11:56Of that, we've spent, you know,
11:58my group has spent a lot of time over
12:00the last however many years to try to
12:02disentangle which of these different
12:04theories might best apply to our data.
12:06And the reason that we've been doing that
12:08is because we think it has important
12:10implications for how we intervene.
12:12So if the data we find are
12:14consistent with an exposure model,
12:16that suggests that we can
12:17intervene at any time,
12:18and it suggests that our goal is to
12:20try to partition the population,
12:22so to speak,
12:23in terms of people who've been exposed
12:25in those who haven't been exposed.
12:26If our data are consistent with
12:28an accumulation model,
12:29then that points us to want to try
12:32to intervene early before people are
12:34accruing those adverse exposures.
12:36If it's a recency model,
12:38it suggests that we want to intervene
12:40quickly,
12:41but it might also suggest that maybe
12:43doing nothing would be OK too.
12:44These symptoms might naturally resolve
12:47or risk would resolve overtime,
12:49and if it's a sensitive period,
12:50model suggests that we want to
12:52intervene during or maybe shortly
12:54before those time periods of
12:57increased sensitivity.
12:57So we think about sensitive periods
12:59as being both high risk periods or
13:02windows of vulnerability when adverse
13:04life experiences are more harmful.
13:06But they could also be windows
13:08of opportunity,
13:09when enriching interventions
13:11could yield greater impact.
13:13So we've been working over the
13:15last several years to try to bring
13:17more research evidence to this.
13:19And my goal ultimately is to try to
13:22enable policymakers and clinicians
13:24to have better data to act,
13:26to know not just what to what
13:28to do to intervene,
13:29but specifically when and to try to do
13:31that on a high resolution timescale.
13:33So in other words,
13:35let's get more granular than just
13:37saying early or saying 1000 days.
13:40The 1st 1000 days rather.
13:43So I want to tell you a little bit
13:45more about the the first work that
13:48we did this was back when I was a
13:51postdoc and was a data set called
13:53AD Health that's now following it
13:55started studying kids when they
13:57were in middle school,
13:58and they've now been following
14:00them through adulthood.
14:02And the way that the data were
14:04recorded in AD health gave us the
14:06chance to ask this question about
14:08sensitive periods because people
14:09were asked were you exposed to
14:11physical abuse and sexual abuse?
14:13And if so,
14:14how old were you when that first happened?
14:17Now,
14:17I know there's measurement challenges here,
14:19and I'm going to come back to that.
14:20But just to give you a sort of
14:22intuition for how we've approached
14:24these sensitive period studies.
14:25So we code people based on whether
14:28they've been exposed to adversity
14:30during these different periods.
14:32And what we end up finding is that
14:34compared to people who were never exposed,
14:37kids who were exposed to physical
14:39abuse generally across the board
14:41have an increased risk of depression.
14:44Compared to kids who are unexposed,
14:46but when we start to compare kids
14:48based on the timing of their exposure,
14:50we do see some within group differences.
14:52So here we see that kids who are
14:54exposed as preschoolers for the first
14:56time have an increased risk of high
14:58depressive symptoms compared to kids
14:59who were exposed for the first time in
15:02adolescence and similarly for sexual abuse.
15:05We find generally across the
15:07board this increased risk,
15:08but here too these potential
15:10sensitive periods this was shifted
15:12to be latency or school age period.
15:14Relative to preschool or relative
15:17to the prepubertal period?
15:20Over the years we've searched
15:22broadly for sensitive periods,
15:24trying to see the level where
15:25they might operate.
15:26So the earliest work that we did
15:28was looking at childhood adversity
15:29in relation to depression and
15:31other forms of psychopathology.
15:33But then we started,
15:34this is really based on my
15:36interest in genetics,
15:37starting to think about these
15:39intermediate phenotypes.
15:40So in other words,
15:42are there these measures that we can
15:44get that are maybe more proximal to risk
15:47based on their timing of occurrence?
15:50In other words,
15:51these measures that are maybe
15:53capturing more of the biology of
15:55the the short-term effects of
15:57exposure to childhood adversity.
15:59And maybe these are the level,
16:01this is the level where we might see
16:03more signal and might be more readily
16:05able to identify sensitive periods.
16:07So we've looked at a number of
16:09different intermediate phenotypes
16:10and I'll tell you more about those.
16:12You know everything from how kids
16:14cope with stress to executive
16:16function and more recently,
16:18molecular targets.
16:21Throughout this work,
16:22we've also been focused on trying to
16:24develop and apply better analytic tools,
16:26particularly to analyze longitudinal
16:28data where you have these repeated
16:30measures of adversity and where
16:32sometimes they're highly correlated.
16:34So this is an approach that we've
16:36been working on with my colleague
16:38Andrew Smith out of the University
16:40of West of England in the UK.
16:42So it's called the structured
16:44life course modeling approach,
16:45or the slick comma and slick
16:48comma is incredibly cool.
16:50It works really well when you
16:53have repeated measures data.
16:55It can work when you have measures that are
16:58measured close in time or more distally,
17:00in time.
17:00What I also really like about it is
17:03that it forces you to have ideas up
17:05front about what you think you might see.
17:08So it's not just a a fishing expedition,
17:10but you have to have some idea
17:12about what you think might be the
17:14theory at play so that you can then
17:17encode your theories into testable
17:19hypotheses. And So what the what?
17:21The way that it works is that it
17:24essentially allows you to identify
17:25from the combination of theories
17:27that you've identified beforehand,
17:29which set explain the most amount
17:31of variation in your outcome.
17:33So it basically works in three stages.
17:35So the first thing you do is you take
17:37all of your theoretical models and
17:39then you encode them into a variable.
17:41So, for example, if you're going
17:43to test a sensitive period model,
17:45you code people based on being exposed
17:48during that time period versus outside.
17:51An accumulation model,
17:52you're coding the number of
17:54exposures and so on and so forth.
17:56And these aren't the only life course models,
17:58I should say, that you could test.
18:00But there's other kinds of models too,
18:02like mobility models.
18:03So where a kid might have social
18:05support and then they don't at the
18:07next time and then it comes back again.
18:10And So what you end up doing is you
18:12take all of these variables and then
18:14you bring them into a regression model.
18:16And the way the regression model is
18:18working is it's in a very sequential
18:19fashion where you're trying to.
18:21Identify the amount of variation
18:23in your outcome or R-squared that
18:25is explained by the greatest
18:28combination of variables and.
18:29So what you can see in this example
18:32is where the model keeps fitting
18:34until it gets to this combination
18:36of both accumulation and exposure
18:38during that third time period.
18:40And So what you're able to also do is
18:42you can look at these elbow plots,
18:44which is what I'm showing here in the middle,
18:46but then you can also evaluate
18:49fit quantitatively using post
18:51selective inference.
18:52And so we've used the slick law
18:54over the years and and also other
18:56analysis to look at you know,
18:58psychopathologies,
18:59suicide risk,
19:00sleep and and other more intermediate
19:04phenotypes.
19:04I want to tell you more about the
19:06work that we've been doing where
19:08we've been engaged in the most
19:10work so far and that's related to
19:13DNA methylation and epigenetics.
19:15So these are chemical tags that are
19:17essentially added to your genome.
19:19They don't change how your DNA sequence.
19:23Is is shaped, but they change.
19:25They have the potential to change
19:27how your genes function.
19:28So they're one pathway through which
19:31adversity might end up affecting
19:33depression and other adverse health outcomes.
19:36So one of the main studies that we've
19:38been using for this is a study called alpac,
19:41where the Avon Longitudinal
19:43Study of Parents and Children,
19:45which for any of you who might
19:46be shopping a data set,
19:47is a wonderful data set they have.
19:50It's a birth cohort and the kids are,
19:52the kids are now in their.
19:5330 So there's you know 30 years of data.
19:57They also collected as part of the
19:59sub sample about 1000 mother child
20:02pairs with DNA methylation data.
20:04And so that was what we ended
20:06up analyzing here.
20:07So we had repeated measures of
20:09exposure to different types
20:10of adversity, things that were happening
20:13within the household up through markers of
20:15neighborhood disadvantage and we coded those
20:17based on the timing of occurrence and then
20:20we looked at these markers of adversity.
20:23In relation to DNA methylation,
20:26a type of epigenetic modification at about
20:30500,000 different sites across the epigenome.
20:33And so we applied the slickman and we asked,
20:35you know, what's the best theoretical
20:37model that might explain the variation
20:40that we see in these epigenetic marks?
20:42Is it accumulation, recency,
20:44sensitive period or maybe a combination?
20:48And what we did was we ended
20:50up analyzing the data.
20:51So on the X axis here is chromosome,
20:54on the Y axis is the negative
20:56log of the P value.
20:57So it's basically the test
21:00of statistical significance.
21:02This is a Manhattan plot.
21:04So ideally it looks like Manhattan
21:05where you see these skyscraper like
21:07effects emerging from the data.
21:09You don't want a Dutch plot,
21:10you don't want it to look flat because
21:13these are basically regions where you're
21:15seeing you know interesting signal.
21:17So and then because we test literally
21:21500,000 different associations,
21:23we correct for that testing.
21:25So anything that's considered to
21:26be above that line is considered
21:29epigenome wide significant.
21:30And so we ended up finding 46 loci
21:33that were distributed throughout
21:35the epigenome as being potentially
21:38impacted by adversity.
21:40And when we dug deeper into these results,
21:43what we ended up finding that
21:45more than half of the loci.
21:47We identified were influenced
21:49by exposure to adversity,
21:50specifically between ages three to five.
21:53I actually didn't expect that we'd find
21:55such strong evidence for sensitive periods.
21:57I was thinking accumulation might
21:59be as important, but it wasn't.
22:01Here we actually didn't identify any loci.
22:04What's also interesting is that
22:06these DNA differences weren't
22:07actually present at birth.
22:08So we looked at whether they happened
22:10in cord blood and they didn't.
22:12And we've been now working.
22:14We're probably about a month off or
22:16so from wrapping up efforts around.
22:17A meta analysis that we've been doing
22:20to try to replicate and extend these
22:22findings and other datasets to see
22:25if they hold and by and large spoiler
22:27alert is I think most of the data,
22:29most of the evidence we are seeing
22:32is for sensitive periods relative
22:34to these other models.
22:36One thing I also want to say too is,
22:39you know,
22:39one of the questions that I think is,
22:41is really fair is these methods
22:43seem really complicated, you know,
22:45is is the juice worth the squeeze
22:47so to speak?
22:48Do you actually get more if you,
22:49you know,
22:50if you get all these repeated
22:52measures and you model it with this
22:54sophisticated modeling approach,
22:55the answer is yes.
22:57So we are able to identify with
22:59the slick ma more signal that we
23:01would have missed had we just
23:03coded people as exposed versus.
23:05Unexposed.
23:06So I think hopefully you hear
23:08a message here of you know
23:10it is worth it to do this
23:12more repeated measures data collection and
23:15use these these more complicated methods.
23:20We've also been working and this
23:22is work led by Alex Lucier,
23:23a postdoc in my group, because we have
23:26longitudinal methylation data in alpac.
23:29So not just looking at methylation at age 7,
23:32but he's also been expanding it
23:34to methylation at age 15 to 17.
23:37So we can understand these patterns
23:39of stability and change across time.
23:41And these, these data are really interesting
23:43and I'm just going to present a a little
23:46bit of what we've been finding here.
23:48So what I'm showing here.
23:49Are the 46 low side that I showed before,
23:53so these were the top low side
23:55that we identified at age 7.
23:57Now we're looking at them at age 15 and
23:59saying do we still see them being you
24:02know largely important and generally
24:04what we find is that the direction of
24:07change or the the pattern of direction
24:11of association is generally the same
24:14but the results are attenuating slightly.
24:16So had we run an epigenome Wide Association
24:19study we wouldn't have identified these.
24:21Game low Sci at age 15.
24:25What we're finding actually now
24:27is a new set of loci at age 15,
24:31so 41 in total and interesting.
24:34These are also underscoring the
24:36importance of this early childhood
24:38of this age three to five period.
24:40So we didn't see them before at 7,
24:42but now we're starting to see them.
24:43So sort of interesting to think about maybe
24:46potential sleeper effects or latency effects.
24:49We don't really know what's
24:50necessarily going on here,
24:51but we're we're starting to
24:53try to unpack this and ask,
24:54you know what might be giving rise to these?
24:56These patterns.
24:57We've also looked,
24:59you know,
25:00at these data and we can find
25:02so far at least six different
25:04patterns of adversity associated
25:06methylation differences across time.
25:08And these patterns essentially
25:10reflect differences that emerge
25:11early versus later in development,
25:14those that happen among people
25:15who are exposed to adversity.
25:17But what's interesting here is we
25:19see differences based on whether
25:21you were exposed during the
25:22sensitive period we think might be
25:24impactful versus outside of it.
25:26And there's some cases where
25:27people who were exposed.
25:29During the sensitive period,
25:31look like people who were unexposed.
25:34And then we're also seeing
25:36differences in in age differences
25:39based on age and assessment.
25:41And I think this is really interesting
25:43in terms of thinking about again
25:44a kind of is the juice worth the
25:46squeeze question of you know,
25:48is it worth us getting these repeated
25:50measures of methylation and and I think
25:52at least from what we're seeing it is.
25:57So. Umm. You might also be wondering,
26:02OK, this is interesting,
26:04adversaries predicting methylation,
26:05but what's actually
26:06happening in terms of health?
26:09And Alex, who's a postdoc,
26:10as I mentioned, and Brooke Smith,
26:12who was a data analyst in my group,
26:14have been doing a mediation analysis,
26:16mediation analysis to essentially
26:18ask is adversity leading to changes
26:21in these DNA methylation signatures
26:23that then predict risk for depression
26:26and what we're looking at here.
26:29So we could basically calculate
26:30all of these different paths.
26:32Using regression and what we're looking
26:34for is to try to identify, you know,
26:37how much of the association is explained
26:40by these methylation signatures.
26:42And So what we found overall is so far
26:4570 total mediators that were identified
26:49across these different adversities,
26:51corresponding to 667 unique CPG
26:55sites that that each explained
26:58between 10 and 71% of the variation.
27:02In risk for depression.
27:04So you can see that there's differences in,
27:06you know,
27:07how much is being explained across
27:09these different types of adversities.
27:11And then what I think is maybe really
27:14interesting is that the epigenetic
27:16adaptation that we're seeing is not uniform.
27:19So when we plot the direction of these
27:22different associations and whether
27:24adversities associated with increased
27:26methylation or decreased methylation,
27:29we're seeing a lot of variation.
27:31So what this is essentially showing
27:33is that most of what we're finding
27:37are effects where methylation changes
27:39are actually protective against.
27:42Depression.
27:43So adversity is associated with a
27:46methylation change that protects
27:48people from developing depression.
27:50We've also been finding that some
27:52sites that we've identified are linked
27:54to cortical development and and other
27:56aspects of brain development and we've
27:58been able to replicate some of the
28:01LOCI and some independent cohorts.
28:03And I think this is another area
28:05that's just ripe for investigation
28:07because it's counterintuitive.
28:09I think most of us would expect
28:11that these things are, you know,
28:13more deleterious.
28:13But it might be that we're,
28:15our bodies are trying to reach homeostasis.
28:17And so we're some of the damage
28:20that's done is protective and
28:22some of it is also harmful.
28:24I also want to just kind of
28:26zoom out and sort of share with
28:28you the last set of work around
28:30sensitive periods in terms of this,
28:32this review paper that John Schaefer,
28:34who's a a postdoc collaborator of
28:37mine and I worked on around the
28:39question of sensitive periods.
28:41So I've been really surprised that the
28:44data we've been seeing for methylation
28:47has been so consistent for sensitive
28:50periods and we wanted to know you know,
28:52does this really extend to.
28:54Other domains.
28:54So we ended up publishing this review.
28:57It just came out a couple weeks ago
28:59looking at a range of different outcomes.
29:02So psychopathology,
29:03neuroimaging, epigenetics,
29:05psychophysiology and behavior.
29:07It's defined by our doc,
29:09the research domain criteria.
29:11So we found 118 unique cross-sectional
29:15observational studies.
29:17Most of these studies focused
29:19on psychopathology as at least
29:21one of their outcomes,
29:22so depressive symptoms or diagnosis
29:25or other PTSD or so on and so forth.
29:30Other ones we're looking at are
29:31other R DOC domains and a handful.
29:33We're also looking at more neural indices.
29:37What we ended up finding was that most
29:39studies did report a timing difference.
29:42In other words, they reported that kids
29:45exposed to maltreatment in one time
29:47period had an increased risk relative
29:48to kids exposed at another time period.
29:51But when we dug deeper into
29:53these timing effects,
29:54we essentially didn't find any consistent
29:57evidence for peak periods of vulnerability.
30:00So it's not as though we saw three to five or
30:046 to 8 is this time period of vulnerability.
30:07This was also very surprising.
30:09So we didn't see that these
30:12biological markers, you know,
30:14the neural indices or other indicators were
30:18any better able than the symptom measures
30:21to identify potential sensitive periods.
30:24We also didn't see any
30:26differences based on study rigor.
30:27So if you had a study where you
30:30compared your models to accumulation
30:32models or you were a larger study,
30:35we didn't see any differences based on that.
30:37Neither we did interestingly share find
30:41that there were similarities in terms of
30:44internalizing and externalizing symptoms.
30:46They did share peak periods of vulnerability,
30:49but specific types of maltreatment did not.
30:51So this maybe speaks to maltreatment
30:54types having potential different impacts
30:57with respect to sensitive periods.
30:59Studies were also split with respect
31:02with respect to sex differences.
31:05We also generally saw a huge risk of bias.
31:09Most of these studies were under powered
31:11and so as a result we ended up providing
31:14a set of recommendations at the end
31:16that we hope will guide future studies,
31:19including but not limited
31:20to issues of of measurement,
31:22which I'm going to turn to next.
31:24So in terms of the issue of measurement,
31:28so this is something I've been
31:30frustrated about for for a while.
31:33So we know that current measures
31:35of childhood adversity have
31:37some pretty serious limitations.
31:39So what we most often do in in research
31:41studies is we ask people and this
31:43is also in clinical practice too.
31:45We ask people retrospectively.
31:47So when you're an adult or maybe an an
31:50adolescent, we ask you how old you know,
31:52did you experience these adverse
31:54events and so.
31:55You might imagine that there's
31:57a lot of potential bias here.
31:58So, you know,
31:59it's subjects to people's memory.
32:01It's subject to whether they're
32:03comfortable disclosing what are
32:05oftentimes very painful events.
32:07So it's no surprise that, you know,
32:09there might be bias in these
32:11retrospective measures.
32:11The other thing that we can also
32:13do is go prospectively.
32:15So we can ask parents,
32:17oftentimes moms,
32:18whether their child is exposed
32:20to certain kinds of events.
32:22But this is another area where
32:24there's potential problems.
32:26So moms might not want to talk
32:28about painful events or events,
32:30particularly when she's the perpetrator
32:33of those sources of adversity.
32:35For adolescence,
32:36there might be some adversities
32:37that parents don't know about,
32:39and I think This is why it's maybe
32:41there's no surprise that when you
32:43ask both children and their parents,
32:45you see very low levels of
32:47agreement between the two of them.
32:49Another source of data would
32:50be the official reports,
32:51like health and Social service records,
32:53but we know that those are also
32:55dramatic undercounts of people's.
32:57Exposure to adversity and they probably
33:00only get about 30% of all true cases.
33:03So, so sort of borne from these
33:05frustrations and a very serendipitous
33:07conversation I had with a colleague
33:10that I started thinking about baby
33:12teeth and this idea that maybe baby
33:14teeth could serve as fossilized records
33:17of people's early life experiences.
33:19So we published this paper.
33:22Back in 2020 in biological psychiatry,
33:24where we outline this hypothesis
33:26and so we said we basically put
33:28forward this teeth conceptual model,
33:31this idea that teeth are as encoding
33:34experiences to transform health.
33:36And So what we said is that
33:37you have this exposure,
33:38so a psychosocial stressor,
33:41it disrupts some biological process.
33:44It leaves behind an imprint of that
33:47biological process somewhere and
33:48that that predicts health outcomes.
33:50And So what we were saying.
33:52That essentially primary tooth
33:53development might be altered as a
33:56result of this adversity and that could
33:59then therefore be captured in baby
34:01teeth that started forming prenatally.
34:04So let me tell you a little
34:06bit more about teeth.
34:07I could talk an entire talk about teeth
34:09because they're like absolutely fascinating,
34:11but in the interest of time,
34:12I won't do that.
34:13But,
34:14but just to give you a little bit more of a
34:16flavor for teeth and in how cool they are.
34:18So,
34:18so most of us are born with 20 primary teeth.
34:21These are our baby.
34:22Death or milk teeth,
34:23they start forming during about the
34:25second trimester of life and then they
34:27continue forming over the first few
34:29years of life and then around age 5 or six,
34:32they fall out.
34:32They're the only part of our
34:34body that actually falls out
34:35as part of a healthy process.
34:37And then they're replaced
34:39by 32 permanent teeth.
34:41And those form postnatally up
34:43through about mid adolescence.
34:45And so teeth are also amazing because
34:47they record the timing of their
34:50incremental growth, so the outside.
34:52Part of our tooth is called the
34:54Crown and that's comprised of the
34:56enamel that we hopefully brush
34:58twice a day in our underlying
35:00dentin and then the pulp and root.
35:02And the way that teeth develop
35:04is really very much reminiscent
35:06of a circadian like process.
35:08So there are cells called ameloblasts
35:10and those are the cells that form
35:12enamel and they're basically acting
35:14in a in a circadian like process
35:16to lay down this matrix of enamel.
35:18And as every sort of passage of time goes on,
35:22it leaves.
35:23Behind an imprint of that recording.
35:26So this is similar to the way that
35:28tree rings develop and that every
35:30year of the trees development
35:31you see a new ring recorded.
35:33Well, our teeth have very similar lines.
35:36There are sets of lines that
35:38correspond to about weekly
35:39development, and then lines that also
35:42correspond to about daily development.
35:44What's also unique is that this recording
35:46of development is found across evolution,
35:48so we see similar tree similar rings
35:51within the teeth across different species.
35:54And teeth also record insults or disruptions
35:56that happen during their development.
35:58So in this way we can think about teeth
36:01just telling us not just whether a stressor
36:03occurred in development but potentially when.
36:05And this can happen on both a low
36:08resolution time scale where you can
36:10see for example these white marks or
36:12these enamel hypoplasia or concentrated
36:14to those two central incisors.
36:16So that maybe speaks to something
36:18that was happening as those particular
36:21teeth were forming,
36:22but then you can get even more granular
36:24and look really at a high resolution.
36:26Time scale and leverage what we know
36:28about those tree ring like structures.
36:30So you could take a tooth,
36:31cut it in half,
36:32take thin sections of the tooth,
36:34put it on a slide,
36:36put it under a microscope and
36:38look at the incremental formation
36:40of that tooth development.
36:42And one of the lines that you can look
36:44at among others is this neonatal line.
36:47So this is a line that actually
36:49differentiates the time of our birth.
36:51So it differentiates prenatal enamel
36:53from post Natal enamel and it's often.
36:56Is in studies of archaeology and
36:58anthropology as as a way of differentiating
37:01those different time periods.
37:03But then there's also other
37:04lines that you can look at,
37:05and these are generally referred
37:06to as stress lines.
37:08Whether they happen prenatally,
37:09anthropologists don't really know.
37:11So this is part of what
37:12we're trying to look at, Umm.
37:14And also these lines occur,
37:17as I said, at different time scales.
37:19So.
37:19So you can get pretty granular with teeth.
37:22Most of the work that's been done
37:24so far around teeth as markers of
37:27stress really focus on Physiology,
37:29physiological stressors,
37:31so disease, malnutrition.
37:33The process of our birth and the recording
37:35of that neonatal line and most of this
37:38happens in archaeological populations
37:39and there is a little bit that's
37:41going on in more modern populations.
37:44But what I think is interesting
37:46is that there are primate studies
37:48that have shown that teeth might
37:50record psychosocial stress.
37:51So Simona Lemmers is a postdoc in my group.
37:54She's a biological anthropologist.
37:56She's been doing some of this work
37:58and essentially shows that different
38:00kinds of events that happen within.
38:03For her study,
38:04it was mandrills.
38:05So you can see these stress lines appearing
38:08shortly after the occurrence of a stressor.
38:11So, for example,
38:13you separate the offspring.
38:14From the mother and you'll see that the
38:18baby tooth will show evidence of that
38:21calendar timed event appearing in the tooth.
38:24So it sort of suggests that maybe there
38:27is something going on about teeth
38:30recording these early life stressors.
38:32So now going from, OK,
38:35so maybe early life stress can
38:37get recorded in teeth.
38:38Well,
38:38what about teeth as a marker
38:40of mental health?
38:40Well, there's been work mostly in
38:43environmental health showing that pesticides,
38:45things that you can ingest or inhale,
38:48those can appear in teeth.
38:50And those are also indicative
38:51of mental health risks.
38:52So for example,
38:54studies have focused on heavy metals
38:57and lead like lead and and other things,
39:01and showing risk for a range of.
39:02Different psychiatric disorders,
39:04autism spectrum disorder,
39:05schizophrenia and the like.
39:08But there really hasn't been a
39:10lot that's been done specifically
39:12in child psychiatry, I think,
39:14and in depression in particular.
39:16And I think this is where teeth Perot
39:18may provide this enormous and unique
39:21opportunity for primary prevention.
39:23So if you think back to the beginning
39:25of the talk where I shared that 20 to
39:2740% of people will have had a first
39:30onset of depression before age 21.
39:31And you think about what I just shared
39:33in terms of the timing of tooth formation
39:36happening early in development,
39:37you can think about every time.
39:39Point when teeth are lost as a
39:41potential opportunity to intervene.
39:43So the first time happens when
39:45teeth are naturally exfoliated,
39:46they fall out of your mouth around
39:48school age, you know, 5 or 6.
39:50Instead of throwing those in the garbage,
39:53what if they were potentially used to
39:55help guide primary prevention efforts?
39:57Similarly,
39:58second opportunity comes for
40:00orthodontic work.
40:01So in the US about 20% of kids will
40:04have at least one tooth extracted
40:06to make room for those braces.
40:08Here too is another opportunity and
40:10also at a time where we start to
40:12see upkicks in risk for depression
40:14and other forms of psychopathology.
40:16And then the last time comes with
40:18wisdom tooth removal surgery.
40:20So this also happens in that transition
40:23to from adolescence to adulthood,
40:25kids are starting to live outside
40:27of the home.
40:27For the first time we start to
40:29see psychosis and other major
40:31psychiatric disorders.
40:32So here imagine again if we might
40:34be able to use these as potential
40:37biomarkers in combination with other
40:39tools to help identify kids at risk.
40:41So so seeing all of these this
40:44under you know under studied area
40:46and seeing the potential I decided
40:48several years ago that I wanted
40:50to become the science tooth fairy
40:52and and try to study baby teeth.
40:55So this is a the cover of a children's.
40:58Look, we actually wrote,
40:59so when kids are recruited into our study,
41:02we use this book as a way of talking
41:04with kids and family about why they
41:06should donate their teeth to us.
41:08I have copies of the book too,
41:09if anyone's interested in it.
41:12I'll just share a very high level,
41:13a couple of ideas part in the pond,
41:15but I just have to. I love puns.
41:17Well, we've been sinking our teeth into,
41:19in terms of this teeth conceptual model.
41:21So Simona Lemmers, Mona lawyer,
41:242 postdocs in my lab and Ryan Lisanne,
41:26who's a pediatric dental resident
41:29at Children's.
41:31So we've been doing work on
41:32the empirical side,
41:33you know, can we see evidence of
41:36markers in in teeth being predicted
41:38by exposure to early life stress?
41:41We published a paper.
41:42Last year showing that markers of Mom
41:45depression and social support were
41:47associated with that neonatal line
41:49and in the direction we expected.
41:51So more stressful births in the
41:54form of higher psychopathology,
41:56wider neonatal line,
41:57conversely more social support,
42:00narrower neonatal line.
42:02And that was also pack.
42:05We also started a study,
42:06we just finished recruitment for
42:08it in the spring called strong
42:09the stories teeth record of
42:11newborn growth where we have.
42:12In recruiting the moms,
42:13recruiting the offspring of women
42:15who are pregnant or raising a
42:17newborn during the timing of
42:19the Boston Marathon bombing.
42:20So the idea here is we have a calendar
42:22dated major stressful life event.
42:24Can we see evidence of that
42:27recorded in kids teeth?
42:29We've also been doing work to then link
42:31what we see in teeth with mental health.
42:33So we published a paper showing
42:35that markers derived from micro
42:37CT of enamel volume and thickness
42:39predicted levels of psychopathology
42:41symptoms in kindergarten age kids.
42:44And then we also have some work
42:46looking at kind of the timing
42:48and pacing of these growth marks
42:50predicting weight gain in adolescence.
42:53And then the last thing that
42:54we've been working on are more
42:55feasibility kinds of studies.
42:56So teeth are new biomarkers,
42:58you know we need to.
42:59You know,
43:00it's scientists and clinicians
43:01how we should be talking about
43:02them with parents and families,
43:03particularly help us understand
43:05how we can use them.
43:06So we've also been doing studies
43:08to try to understand.
43:09What do people think about teeth?
43:12I I laugh when I get emails about them.
43:14Sometimes Mom will say, you know,
43:16dear Doctor Dunn, I've read about your study.
43:19I saved my child's baby teeth.
43:20And then it's either one of two things.
43:23I'm so glad I saved them or you.
43:25Isn't that gross?
43:26I don't know why I saved them.
43:27Something along those lines.
43:29So.
43:29I think there's a lot that we
43:31can potentially learn here,
43:31and I think that will be important
43:33for building a solid foundation
43:35for this work to unfold.
43:37So let me just wrap up by saying a
43:39little bit more on the translational
43:41side in terms of where I see this
43:44work potentially going in terms of
43:46promoting resilience and trying
43:48to reduce health disparities.
43:50You know we talked very early on
43:52in the pandemic about us living
43:54through unprecedented times.
43:55I don't feel like we,
43:56you hear that as much now,
43:57but I still think we very much are.
44:00So you know,
44:01we have all these stressors that
44:02people experience before the pandemic,
44:05you know add on these additional stressors.
44:07That people are experiencing as
44:09a result of the pandemic.
44:10But I also think simultaneously
44:12we're seeing these shifts that are
44:15happening largely as a result of the
44:17civil rights movement around racial
44:19equality and some institutional
44:21practices that are
44:22starting to shift where.
44:27Oh, OK. Thank you.
44:29Umm, where we're seeing some
44:31movement to potentially better
44:33address some of these areas.
44:34And I think where we are as scientists
44:36and also as clinicians is that,
44:38you know, we have the chance to really,
44:40I think, develop a deeper and more meaningful
44:43research agenda to try to understand,
44:45you know, opportunities to identify
44:48ways to promote health and reduce
44:51risk and build some interventions
44:53to really promote resilience.
44:56I think there's at least two.
44:57Main starting points for
44:58where we can go in this front,
45:01I think one is we spend a ton
45:03of time focusing on the bad.
45:05We do a lot of work and adversity and trauma,
45:08you know,
45:09and I think the resilience world,
45:10there's definitely been a
45:11lot of work in this area,
45:13but I don't think that the
45:15resilience work has necessarily
45:16been as integrated in areas of
45:18biology where I think it could.
45:20So I was sharing with some of you that
45:22we just had a grant that hopefully
45:24will get funded that will allow us
45:26to look at the biological embedding.
45:28Of protective factors.
45:29And I think this is something that
45:31we need to bring in as part of our
45:33research model so that we're not just
45:35studying risk because we know that
45:37risk alone doesn't predict outcomes,
45:39but it's a constellation
45:41of different factors.
45:42I think the other thing too is that
45:45we also need to do more to develop
45:47and implement tools to measure
45:49childhood adversity and differentiate
45:51exposure from the biological
45:53consequences of that exposure.
45:54And I think this is really, really hard,
45:57but I'm hoping maybe we're baby.
45:59Keith and and some of our epigenetic
46:01work can go and I think this is really
46:03critical because you might find that
46:05some kid has the exposure but seems
46:07to be doing OK you know there's
46:09individual differences in this adversities,
46:11not deterministic.
46:12So I think being able to disentangle
46:15these is going to be really critical.
46:17In terms of the applications and
46:20implications of the epigenetics
46:21and exfoliated primary teeth work,
46:24you know, I'm an epidemiologist,
46:25so I don't always have the the
46:27the fortune of being able to talk
46:30with parents and families.
46:31But when I do,
46:32I'm always struck by the questions they ask.
46:35And they always ask two things.
46:37The first thing is they want answers.
46:38They want to know why did my loved
46:41one develop a mental health issue?
46:43They want to know if you know their
46:45child being exposed at this age,
46:46you know, caused this,
46:47and then they also want hope.
46:49They want to know what can be done
46:50to prevent some mental health
46:52issue and someone else they love.
46:54And so I think, you know,
46:55what if baby teeth,
46:57when paired with existing tools
46:58and insights like family history
47:00and genetic and other markers,
47:02could provide answers to some
47:04of these burning questions.
47:06And you know,
47:07they these things are something
47:09that naturally
47:09fall out of our mouth and most times they're
47:12either stored or they're thrown away.
47:14But what if instead, these really hidden
47:16in plain sight objects could be used
47:19to give new insights that could help
47:21identify people that might be at risk,
47:23and use the data from that to target
47:27towards specific strategies for prevention?
47:30And I think this is where maybe one day
47:33we might be able to add methylation
47:36signatures or these epigenetic signatures
47:38and teeth as part of our screening tools.
47:41So, you know, imagine a world where somewhere
47:43in the future a child loses a tooth,
47:45whether it falls out,
47:47it's lost for orthodontia
47:49or wisdom tooth surgery.
47:51And that tooth is taken to a healthcare
47:54provider who sends it off then to
47:57a specialized lab and that that
47:59lab is then able to combine.
48:01Data from other omic markers,
48:03genetic markers and epigenetic markers
48:05and survey data about early life
48:08stress and other stressors and more
48:10about the family context and pair that
48:13with family history data and that you
48:15could then use that to then identify
48:17people who might be at highest risk and
48:20connect them with preventative treatments.
48:22I think there's a lot we have
48:23to do on this space,
48:24but I think it's really promising when
48:26we think about what we know already,
48:28we know that exercise is protective,
48:30we know that social support.
48:31Protective.
48:31So can we get that data in the hands
48:34of people and create interventions
48:35that really leverage that so that
48:38we can try to reduce risk?
48:39And it might also be someday too
48:41that we're able to shift these
48:43methylation signatures too.
48:44So we see something turning on
48:45that might be deleterious,
48:47maybe there's an intervention,
48:49biological or not,
48:51that can also produce those shifts.
48:53And then in just my last slide,
48:55I'll also say too that I think one thing
48:58that we also want to be mindful of IS,
49:01is.
49:01This idea of of screening and I
49:03think there's a lot of interest
49:06in people doing screening.
49:07I think we have to be careful around
49:10screening though and and we published this,
49:12this commentary a couple months
49:15ago where we tried to just put a
49:17little bit of context around this
49:18area of screening because I think
49:20we're at this tipping point where
49:22there's the potential,
49:23the real potential for screening for
49:25childhood adversity to do potential
49:27more harm than it does good.
49:29So in this commentary.
49:30We just described some recommendations
49:32that folks should consider when
49:34deploying these kinds of screenings.
49:36And you know,
49:37being very clear about what things
49:39measure and and deploying screening
49:41at the right time and making sure that
49:43there are appropriate interventions to use.
49:45And also just creating systems that
49:47are nimble and adaptable knowing
49:49that the science of adversity and
49:51resilience is changing and therefore
49:52we want to be able to leverage
49:54that best evidence in support of
49:56of future interventions.
49:57So with that, just to thank everyone who's.
50:01And part of my career journey
50:05and my collaboration team.
50:07Immigration is good because
50:09science is a global enterprise.
50:11And thank my outstanding lab
50:14members and sources of funding and
50:16I'm happy to take any questions
50:18you might have. Thank you.
50:26Wonderful. Thank you so much Doctor
50:28Dunn and fantastic mix of topics there.
50:31And I know that we've already
50:32got some questions on the chat.
50:33Are there any questions in
50:33the room to get us started?
50:42And so one question that we had in
50:44the chat and actually from Doctor
50:45Martin was can you talk about the
50:47parallels between telomere length and
50:49some of those markers that you're
50:51observing in teeth? Oh, there is.
50:58I thought you were maybe going to ask
51:00about parallels between Umm Tillman
51:02or length and epigenetic aging.
51:05We don't. What are you? So. Umm.
51:09So I think that this is an area
51:12where I don't know that I've seen
51:15a lot of very good comparisons.
51:18There's the epigenetic clocks
51:19that people tend to use.
51:21There's now, there's now about
51:231/2 a dozen dozen of them.
51:26Some of them are correlating with each other,
51:28some of them aren't.
51:29It depends on what tissue type you get,
51:31whether you have buckle cells
51:33or saliva or blood.
51:34So I think part of what we as a
51:37field have to grapple with is trying
51:39to build studies that allow us to
51:42better understand similarities and
51:43differences in these markers and then
51:46also piece together that with the context of.
51:48Development because as I shared,
51:50a lot of these markers also vary,
51:52you know,
51:53over over the course of lifespan,
51:55telomeres and teeth,
51:56I haven't thought about it and we
51:59haven't done anything on that just yet.
52:01I think teeth are really understudied
52:03and an area where there's a lot
52:05of a lot that we can learn.
52:07I don't know if there maybe there's
52:10something in that circadian process
52:13that can be indicative of of aging
52:16related processes or something,
52:18but I have we haven't gotten.
52:20To that yet, but but it's a great question,
52:22something to think about.
52:30I just had a quick question about.
52:33The you brought up measure difficulties with
52:36measurement and bringing it back to age,
52:38and age being kind of just
52:40a proxy for development,
52:41and then you're interested in
52:43looking for sensitive periods.
52:47I noticed in across the development
52:50age was bent in and I think routinely
52:54about two year increments and I'm
52:57wondering if that is was informed
52:59by by research or because that
53:02really can either hinder or help.
53:04Finding these sort of sensitive periods,
53:08if something falls in between
53:09one of those bins or so I just
53:11wondering if you could speak to
53:13how those are are gathered.
53:15I love your question and doing
53:17sensitive period work in relying on
53:19age I think as your questions may be
53:22saying is just an imperfect measure.
53:24So we we tend to use the most.
53:29The narrowest age we can and then we
53:33afterwards Bennett into developmental stages.
53:36So in other words we try to leverage.
53:38So we have differences based on month of age.
53:41So we have eight months and you know 17
53:44months or whatever and then we'll group
53:45after we do the analysis into just a
53:48developmental stage and the thinking there
53:50is that's just sort of how we think,
53:52we think of you know based on school
53:54age and non school age or preschool
53:57period or what have you so.
53:59It's really just meant to try to
54:01help better translate that work.
54:03I think really to understand
54:05sensitive periods,
54:05we need to have measures of plasticity.
54:08And in order to have measures of plasticity,
54:10we need to know what plasticity
54:12actually is and how what we mean
54:14by it and how we define it.
54:16So I have a postdoc in my group
54:18that's actually working on that.
54:19That's just saying can we get all on the
54:21same page about what we mean by plasticity.
54:24So the plan is to write a paper on that
54:26and then to follow that with a paper on,
54:28OK, now that we're hopefully maybe.
54:30More on the same page about plasticity.
54:32Can we then start to think
54:35about markers of plasticity?
54:36Because we're a lot of us are
54:38really interested in plasticity.
54:39But plasticity means something really
54:41different to a neuroscientist who
54:43thinks about it at a synaptic level and
54:46someone who's thinking about it in the
54:48context of like stroke recovery for example,
54:51and and rehabilitation and
54:52those kinds of outcomes.
54:54So I think this is another,
54:56I think this is a Holy Grail
54:58for our field is to really,
54:59I think if we nail the sensitive.
55:01Period question and we did that through
55:03plasticity and had a good markers of that.
55:05I think that would be pretty,
55:06pretty amazing.
55:07Thank you.
55:12Aye, thank you for such an amazing talk,
55:16so I'm very curious about.
55:19The effects that you observe on
55:22the protective effects of the DNA
55:25methylation changes that specific loci.
55:28Could you elaborate a little bit
55:30more how they were defined and also?
55:32Will that be dependent depending on?
55:36That it occurred during the sensitive
55:39periods and that you like look at that
55:42during that specific time whereas.
55:44You know, compared to auto hold for example.
55:46How would that compare? Um, like?
55:50I will argue that maybe adulthood we
55:53will observe more deleterious effects.
55:56But you know, I wonder about.
55:59Your thoughts on that and I
56:00have a second question,
56:01but if you want to answer that,
56:03thank you for just asking one at a time.
56:05That's that's great.
56:07I think so.
56:08I think there's a lot to still unpack
56:10in this mediation work and there's not
56:12been from what we've seen any other
56:14work that's been done in this space.
56:17And it took a lot for us to
56:19just figure out the methods,
56:20you know,
56:21because you're bringing these
56:22methods that have been developed
56:24typically for what we call like a
56:26small data setting and then you're
56:27applying it to data where you have,
56:29you know, as you know,
56:31500,000 different associations that you can,
56:34you know, study.
56:34So a lot of the time we spent,
56:37you know,
56:37was was built in on that and then we
56:40carried forward our sensitive period
56:42work to try to bring in more of this
56:45information about these different
56:46life course models to try to.
56:48Understand, you know these effects.
56:51I don't.
56:51I was not expecting to see such variation.
56:55I think the next set of questions
56:57that will be really key is, you know,
56:59we just looked at depression at one
57:01point in time in late adolescence.
57:03We can look at later markers of
57:05depression to see if this persists.
57:08And I think, you know,
57:10I come from the camp of let's see it
57:12once and if we see something interesting,
57:14let's try to see it again in
57:16another data set and try to.
57:18Replicate it.
57:19And then that's where let's if we do that,
57:22then let's start digging in on biology.
57:24Let's get into cell culture models,
57:26let's get into animal models.
57:28Let's, you know,
57:29really try to probe this to see if this is,
57:32you know, real and what might be
57:35some of the the consequences.
57:37Thank you.
57:38And then make my second question is sort
57:40of a follow-up of their previous question,
57:43how these sensitive fears
57:44are defined in terms of?
57:48The age or and following up on on
57:53the definition of of that in terms of
57:56like have you considered biological
57:58age like not only tell me your land
58:01but also epigenetic aging and I
58:04always wonder what does that mean
58:06during childhood because we often see
58:09accelerated at beginning of aging in
58:11adults being associated with trauma
58:13but that what does that really mean.
58:17In childhood and if these could be?
58:20A marker for biological aging to
58:23define better sensitive periods.
58:25Yeah, that's a great question.
58:27So believe it or not,
58:28I think it's counterintuitive in a way
58:30for us to think about kids as aging,
58:32but they are.
58:33And we, we did a study actually in
58:36alspach where we showed that some
58:38early life markers of stress were
58:41associated with accelerated aging at
58:43age 7 by as much as seven months.
58:45So a 7 year old could look,
58:48you know, Cellularly older than us.
58:507 year old by they would look 7 point you
58:53know seven years with seven months added on.
58:56So I think for us we wanted to
59:00have in order to do the sensitive
59:02period work you need to either have.
59:05You ideally have repeated measures so
59:08that you can get these markers of timing,
59:12not that you're relying on
59:14retrospective reports.
59:15So we there's not a lot of data sets
59:17that have that repeated methylation data
59:19where you could derive those repeated scores.
59:22But we just got a grant last year where
59:25we're doing work in a South African
59:27cohort and where we're going to have,
59:29we're going to be driving epigenetic
59:32signatures at 1/3 and five and I think
59:34that would be a. Great opportunity.
59:36I'm glad you said this.
59:38I think this is something we should
59:40look into there and see if if how
59:42similar or different it is relative
59:43to findings you get for age.
59:46So I know we're almost at time,
59:48but I didn't realize Doctor Lombroso
59:49that your hand was up actually was
59:51fading into the background there.
59:52So Paul, please, please ask your question.
59:56So can you hear me?
59:58Yes. Yes. OK great.
59:59I I that was a fantastic talk and
01:00:02then specifically because it was
01:00:05introducing such a for me anyway a
01:00:07novel area couple of questions that
01:00:10I try to get my head around this.
01:00:13If a child has early onset
01:00:15depression or childhood psychosis
01:00:20or early childhood onset diabetes,
01:00:24are you saying that that there will be a
01:00:26marker for for in in the teeth of this event.
01:00:30And are the epigenetic.
01:00:33Findings,
01:00:35I would imagine they're all different
01:00:38in these three very distinct disorders.
01:00:41Just to help me understand,
01:00:42you probably already mentioned this, but
01:00:45no, I think it's a good,
01:00:47I think it's a good question.
01:00:48Pardon me, I don't know.
01:00:49I think I'm going to look
01:00:51here even though you're here.
01:00:53Trying to answer you, but Umm,
01:00:56so in terms of what we've seen so
01:00:58far with teeth and psychopathology.
01:01:01So we are correlating marker that's
01:01:04derived from micro CT imaging about
01:01:06how thick the enamel basically marker
01:01:09of enamel volume is and seeing that
01:01:12kids who have thinner volume have
01:01:15higher psychopathology symptoms.
01:01:16I think that's just correlational.
01:01:19Who knows whether this is actually causal.
01:01:22I think we need more studies to try to.
01:01:24Impact that.
01:01:25I think teeth might be a marker.
01:01:28So I think of teeth as the marker
01:01:30of early life stress and then
01:01:33potentially those biological
01:01:34markers of early life stress can
01:01:37be informative for mental health.
01:01:39I don't know that teeth necessarily
01:01:42independent of early life stress would
01:01:44be informative for mental health.
01:01:47However,
01:01:47I think there is maybe a kind
01:01:50of tooth brain access where
01:01:52teeth might be informative.
01:01:54For characterizing and understanding
01:01:57processes of brain development that might
01:02:00be harder to interrogate otherwise.
01:02:03So that's sort of my thought on on that.
01:02:06And then whether the epigenetic
01:02:09signatures are similar across disorders,
01:02:11we've not really looked at that,
01:02:15but I think it's something that
01:02:17that you know the work that has
01:02:19been done is more so where you
01:02:21group kids into internalizing
01:02:23versus externalizing symptoms.
01:02:25Part of the challenge is that you
01:02:28really just need incredibly large
01:02:30sample sizes in order to find potential
01:02:33signal when you bring epigenetic work
01:02:36to psychiatric disorders on the order of,
01:02:39you know, thousands, 10s of thousands.
01:02:42To give you context,
01:02:44you may or may not know about
01:02:47genetic association studies.
01:02:48Those are starting to see results.
01:02:51After 500,000, you know,
01:02:53a million participants,
01:02:54so.
01:02:55I think we're seeing more with epigenetic
01:02:57work starting to emerge with less than that,
01:03:00but it's still the scale is very,
01:03:02very large because these effects
01:03:04are are pretty small.
01:03:09Great. Well, thank you all for
01:03:10this rich discussion and please
01:03:12join me again in thanking Dr Dunn
01:03:13for a wonderful presentation.