YCSCAN2 April 2025 Webinar

May 01, 2025

Alongside increasingly permissive social attitudes and laws, cannabis use has increased, especially among pregnant women. However, little is known about potential adverse outcomes associated with prenatal cannabis exposure. Using data from the ongoing Adolescent Brain and Cognitive Development (ABCD) Study (n=11,489), we find that prenatal cannabis use is associated with broad spectrum childhood psychopathology, early cannabis use initiation, and variability in brain, and in particular white matter, structure. Aprils 2025's webinar featuring Dr. Ryan Bogdan discusses findings regarding the ABCD study.

Information

ID: 13087
To Cite: DCA Citation Guide

Credits

Deepak D'Souza

Download Transcript

00:02Good afternoon, everyone,
00:04and I'm Cyril D'Souza.
00:06It's my pleasure to,
00:09to introduce,
00:10Ryan Bogdan,
00:13who is the Dean's distinguished
00:14professor of psychological and brain
00:16sciences at WashU
00:18in Saint Louis.
00:20His research examines how genetic
00:22variation
00:23and environmental
00:24experience
00:26contribute to individual differences
00:28in brain function,
00:30behavior and psychopathology.
00:33He's used a variety of
00:34measures, including
00:36molecular
00:37genetics, GWAS,
00:39MRI,
00:40EEG,
00:41twin studies,
00:43behavioral assessments,
00:45endocrine and information assays, and
00:47self report in both
00:49healthy individuals and clinical populations
00:52across
00:53the lifespan.
00:55His, impactful work has has,
00:58been recognized in many different
01:00ways. He was honored as
01:01a fellow
01:02of both the American Psychological
01:05Association, the APA,
01:06and the Association of Psychological
01:08American
01:10Association for Psychological Sciences, APS.
01:14His work has been published
01:15in many high
01:17impact journals and relevant to
01:19to the mission of our,
01:21the Yale Center for the
01:22Science of Cannabis and Cannabinoids.
01:24He's gonna touch upon a
01:26really important topic about,
01:29how exposure to cannabis,
01:31during fetal brain development,
01:33what impact it has on
01:35behavior,
01:36and and brain,
01:38structure and function. So
01:40that, I'd like to welcome
01:41Ryan. Ryan, great that you
01:43were able to make.
01:45Yeah. Well, thank you all
01:46so much for for having
01:47me. Can everybody see the
01:48screen okay?
01:49Yes. Awesome. I'm gonna go
01:51ahead and get started and
01:52talk about,
01:53prenatal cannabis exposure and its
01:55association with some childhood
01:57outcomes in both some older
01:58samples as well as some
01:59neonatal and infant samples today.
02:02I don't have any disclosures,
02:04to report here.
02:07And just to outline things,
02:08I'm gonna focus on two
02:09studies today mostly.
02:11One is the ABCD study
02:13or adolescent brain and cognitive
02:14development study. So this is
02:15a large national effort that
02:17recruited nearly twelve thousand kids
02:19when they were age ten,
02:20and there's a bunch of
02:21retrospective reports of prenatal cannabis
02:23exposure. So we'll we'll be
02:24talking about that primarily.
02:26But then we also have
02:27a local study that we're
02:28working on here called Cuddle,
02:30or the cannabis use during
02:31development in early life,
02:33in which we recruited,
02:35pregnant individuals during their first
02:36trimester and then follow them
02:37through their pregnancy. We're doing
02:39neonatal brains images on them
02:40and then following them up
02:41through infancy,
02:43with behavior and neuroimaging. So,
02:46we're just getting completed with
02:47our six month data there.
02:48So I don't really we
02:50won't have behavioral data there
02:51for you, but we have
02:52done some stuff tracking, cannabis
02:54use during pregnancy there. And
02:55then lastly, I'll be ending
02:56with some future directions about
02:58some work we're doing with,
02:59Cuddl as well as the
03:01healthy, brain and child development
03:02study, which is another national
03:04effort where we're recruiting,
03:06over seven thousand pregnant individuals
03:07and following their children.
03:10So before I do get
03:10started, I do wanna,
03:12acknowledge some collaborators.
03:13Been able to work with
03:14some, wonderful folk who are
03:16now starting their own labs
03:17at at various places who
03:18contributed and led much of
03:19this work.
03:20And then my colleague Cynthia
03:22Rogers, who I work on
03:23with the cuddle study as
03:24well as HPCD,
03:26and, my collaborator and wife,
03:27Arpana Agrawal.
03:29And then I just wanna
03:30also acknowledge that we
03:32work a lot with ABCD
03:33study data. So this is
03:34that large national effort that
03:35a lot of people have
03:36put a lot of time
03:37and effort into collecting and
03:38and,
03:39and managing those data. And
03:41then, I'll also be talking
03:43at the end about this
03:44HBCD project, which is kind
03:46of inspired by ABCD, and
03:47we're just trying to shift
03:48the the clock back a
03:49little bit further to look
03:50at earlier development.
03:52So as I'm sure you
03:53all are quite well aware,
03:55there have been drastic changes
03:56in,
03:57kind of the social cultural
03:58permissiveness
03:59and laws surrounding cannabis. Cannabis.
04:01And this can be seen
04:02even in perceived political liability,
04:04right, in the the nineties
04:05from, Clinton's original statements versus
04:07where it was seen as
04:08a liability to have tried
04:09cannabis,
04:10politically, but then it was
04:11seen as a liability to
04:12not have,
04:14used it later,
04:15in his political career. And
04:17there's no mere changes that
04:18you see in cannabis use.
04:19So what you're looking at
04:20over on the right
04:21are data from the monitoring
04:22the future study, which tracks,
04:24adults as well as as,
04:27children,
04:28and high schoolers, for example,
04:30looking at daily use. And,
04:31essentially, over the past couple
04:32of decades, there has been
04:33a steady increase,
04:35in cannabis use both from
04:36a, you know, a past
04:37kind of kind of year
04:38use pattern that almost, you
04:40know, approaches fifty percent in,
04:42young adults.
04:44And daily use has also
04:45nearly doubled in this time,
04:47about to seven and a
04:48half percent now.
04:49And this is seen across
04:50groups. We're actually seeing increases
04:52in cannabis use among older
04:53adults.
04:54And the other area we're
04:55seeing a lot is pregnant
04:57individuals.
04:58So in the top right
04:59is,
05:00data from Nora Volkow's
05:02group using the NASDA data
05:04that is showing that essentially
05:06in the last
05:07decade and a half up
05:08to twenty seventeen,
05:11cannabis use among pregnant women
05:13essentially doubled. It went from
05:14about three and a half
05:15percent to seven
05:17percent. A lot of that
05:17use has been in the
05:18first trimester, but it has
05:20it does seem to continue
05:21throughout, and daily use has
05:23also about doubled during this
05:25time during pregnancy.
05:26And there are a host
05:27of reasons for this. A
05:28lot of individuals report using
05:30it for nausea,
05:31especially during the the first
05:33trimester or using it for
05:34pain or discomfort or to,
05:35you know, for perceptions that
05:37it
05:38enhances their sleep as well.
05:41But, you know, the the
05:43outcomes associated with it are
05:44are really unclear.
05:45I do wanna emphasize that
05:46the there are some data
05:47to suggest these might also
05:48be underestimates. There are data
05:50from Kelly Young Wolf's group
05:51and then the Kaiser Permanente,
05:53dataset in California
05:54that is showing rates as
05:55high as ten percent.
05:57And then lastly, the HPCD
05:59study, which is ongoing with
06:00recruitment, and, again, we're we're
06:01trying to get to, you
06:02know, over seven thousand pregnant
06:04individuals.
06:04But in that dataset, we're,
06:06preferentially recruiting for substance exposure,
06:08so it's a little biased.
06:09In particular, we're trying to
06:10actively seek out opioid exposures
06:12in this this group. But
06:13in that dataset currently,
06:16the most used substance is
06:17cannabis, which you can see
06:18here in in red. And
06:19the other interesting thing to
06:20note is that, so these
06:22little dots here represent co
06:23occurring use or individual use,
06:26is oftentimes cannabis is being
06:27used in isolation during pregnancy.
06:30But I think it's quite
06:31striking that it's being used
06:32more frequently,
06:33than alcohol,
06:35now as well during during
06:36pregnancy in this in this
06:37dataset.
06:40There are a host of
06:40reasons for this, but there
06:41are these perceptions that is
06:43this natural substance and therefore
06:45safe, for example. So the
06:46New York Times had this
06:47really neat kind of, anecdotal
06:49reports where they talked to
06:50women, and they talked about
06:51how they're being very careful.
06:52They're only using marijuana. Right?
06:54They're not using alcohol. They're
06:55not using these other things,
06:56these perceptions of safety.
06:58But what is concerning is
07:00that we know remarkably little
07:01about cannabis use during pregnancy
07:03and its potential impact on
07:04offspring. And as I'm sure
07:06you all are well aware,
07:07given the endocannabinoid system's,
07:09involvement in early neurodevelopment, axonal
07:12guidance, neurogenesis, all of those
07:14sorts of things, synaptic plasticity,
07:16it could be a developmental
07:17period that could be associated
07:19with, increased risk.
07:22The one thing that I
07:23think is relatively clear in
07:25the literature, at least according
07:26to meta analyses,
07:28is adverse associations with birth
07:30outcomes among
07:32pregnancies exposed to cannabis.
07:35And this is across low
07:36birth weight. You find generally
07:37find, individuals exposed to cannabis
07:40are at a lower birth
07:40weight, are smaller for gestational
07:42age, more likely to be
07:43a preterm delivery or a
07:44NICU admission across studies. So
07:46that's what you see across
07:47these. And it's about almost
07:48a twofold odd ratio increase,
07:50so almost doubling the likelihood
07:52of those things.
07:53The one thing I do
07:54wanna note is some of
07:55these studies,
07:56have not very been able
07:57some of them have been
07:58able to really well control
07:59for other substance exposures,
08:01Others have have not been.
08:02So the extent that some
08:03of these data reflect,
08:05you know, an aspect of
08:07cannabis or possibly co occurring
08:08environmental circumstances,
08:10nicotine and copamine positivity and
08:12the like,
08:13does remain unclear. But this
08:14is where the data are
08:15probably the strongest,
08:17with our national academy,
08:19you know, kind of declaring
08:20that there there was sufficient
08:21evidence to, you know, proclaim
08:23this negative,
08:24association at least.
08:27When you look at child
08:28behavior in biology,
08:30the findings are incredibly inconsistent,
08:33and they've largely relied on
08:34three main studies like the
08:36Ottawa pregnancy study that have
08:38tracked very small samples of
08:40cases and controls. And a
08:41lot of those cases and
08:42controls are
08:43tremendously
08:44confounded,
08:46by, you know, factors such
08:47as socioeconomic status and and
08:49and the like where they're
08:50they're quite different. So and
08:52across these studies, there have
08:54been very inconsistent findings even
08:55within studies. Sometimes at a
08:57one h, they'll find, you
08:58know, cannabis use is associated
09:00with differences in cognition or
09:01lower cognition.
09:02In another study, they'll a
09:04few years later, they'll find
09:05no differences, and then they'll
09:06also find increases in other
09:08samples. So it's been incredibly
09:10conflicting,
09:11perhaps through these small samples
09:12and and limited limited control.
09:16So in the present set
09:17of studies I wanna show
09:18share with you all, we
09:20were kind of interested in
09:21just examining what prenatal cannabis
09:22exposure is associated with with
09:24regard to childhood behavior and
09:26biology in childhood.
09:29And apologies. I'm seeing some
09:30things in the chat. I'm
09:32terrible with tracking Zoom while,
09:34presenting. So if anybody wants
09:35to, you know, talk about
09:36anything, please just interrupt me,
09:38so I don't have to
09:39track the the chat as
09:39I'm trying to present.
09:42So in the first study
09:43I wanna share with you,
09:44we relied on the ABCD
09:45study. So this is a
09:46large generalist study that everybody
09:48has access to. So if
09:49you're interested in child development,
09:51and you're not using this
09:52dataset, I'd encourage you to
09:53consider it. It's a study
09:54of almost twelve thousand children
09:56that were recruited around, the
09:58ages of nine to ten,
10:00kind of in the the,
10:02like,
10:03a twenty sixteen ish on
10:04a twenty twenty ish. It
10:06is relatively nationally representative or
10:08as kind of as close
10:09to that as studies get.
10:11There are twenty one, twenty
10:12two sites across the United
10:13States. One of the sites
10:14moved, which is why we
10:15often have to consider twenty
10:16two sites.
10:17But it also has a
10:18longitudinal family based component.
10:20So there are a couple
10:21of sites, that recruited monozygotic
10:23and dizygotic twins, and we're
10:24finally starting to get sufficient
10:25power to start to look
10:26at some of, that discordancy
10:28as well.
10:29But essentially every couple of
10:30years, there's neuroimaging.
10:32Every year, there's annual in
10:33person behavioral assessment. And then
10:35in midyear assessments, there's even
10:37phone interviews looking at substance
10:38use and the like. So
10:39it's a really rich, relatively
10:41deeply phenotyped dataset,
10:44given its its size in
10:45particular. And there are some
10:46biomarkers and genetic data associated
10:49with it as well.
10:51So we were interested in
10:52looking at prenatal cannabis exposure.
10:54So this is just showing
10:54some of these differences.
10:56So this is all based
10:57on retrospective report of the
10:58caregiver, which is mostly the
11:00the mother that was reporting,
11:02although that there were about,
11:03like, eight percent of individuals
11:04who were not the the
11:05biological mother,
11:09person that's to the study,
11:10but they were asking about
11:11the maternal's use. And,
11:14this does kind of mirror,
11:16for the most part, national
11:17datasets with regard to, the
11:19percentage of use where about
11:21six percent of individuals are
11:22reporting use.
11:23And there are some, you
11:24know, important demographic differences across
11:26these groups. You find the
11:27greater proponents of
11:29black individuals, for example,
11:31reporting use.
11:33You also find lower, socio
11:35economic status associated with use.
11:38And we did divide this
11:39up into preknowledge exposure and
11:41postknowledge exposure.
11:42And some of the reason
11:43for that is that this
11:45actually,
11:45when women on average when
11:47they were pregnant in this
11:48study was around six weeks,
11:50which is right around the
11:51time according to, you know,
11:52nonhuman animal data that the
11:54endocannabinoid receptors in the brain
11:55developed. So we were hypothesizing
11:57that that would be the
11:58only way that we mechanistically
11:59see this. Although, as I'm
12:01sure you're also well aware,
12:03endocannabinoids
12:04play a huge role in
12:04the intrauterine environment and particularly
12:06with the placenta. So I
12:07don't wanna be minimizing that.
12:09I think there could also
12:09be very, you know, indirect
12:11pathways as well. But in
12:12our analysis I'm gonna present
12:13to you today, we covariate
12:14for things like birth weights
12:15and the like to try
12:16to account for those, you
12:18know, other possible mechanisms. But,
12:19I think they are very
12:21interesting ones as well that
12:22we're we're pursuing in some
12:23future work.
12:25So one other thing that
12:26I think is is really
12:27key about this study is
12:29that we do have a
12:30host of other measures. So
12:31unlike these other, you know,
12:33epidemiological
12:34associative studies that have been
12:35conducted, we can perhaps really
12:37well account for confounds.
12:39And I think this is
12:39one of my favorite studies.
12:40It's an older study, but
12:41I think it nicely exemplifies
12:42why this is important.
12:44This is a study using
12:45some European,
12:46health data,
12:47and essentially what they're looking
12:49at is smoking during pregnancy.
12:50This is looking at nicotine
12:51use during pregnancy.
12:52And one of the things
12:53that has been consistently found
12:54with this is that smoking
12:55during pregnancy is associated with
12:58greater externalizing behavior in children.
13:00What you're looking at here
13:01are nonviolent criminal offenses.
13:04And up on the top
13:05the top of the panel
13:06a here is just a
13:07population based comparison, where you're
13:09looking at individuals who are
13:11smoke who who are born
13:12to pregnancies in which smoking
13:13occurred by the solid line
13:15as well as pregnancies where
13:16there's no smoking in the
13:18dashed line. And as you
13:19can see here, there's a
13:20vastly increased risk
13:23among
13:24children,
13:25who are and this is
13:27as they're adults, essentially, entering
13:29young adulthood,
13:30who are exposed to smoking
13:31during pregnancy.
13:33But the key thing that
13:34they did was in this
13:35dataset,
13:37there was a substantively large
13:39number of siblings,
13:40and there were siblings that
13:41had discordant pregnancies, where one
13:43sibling was exposed to nicotine
13:45use, while while they were
13:47gestating, and the other one
13:48was not. So it's a
13:49really nice discordant design. Of
13:50course, there could be, you
13:51know, other factors associated with
13:53why, you know,
13:54nicotine was being used in
13:55those pregnancies or not. So,
13:56you know, there are certainly
13:57compounds that could could still
13:59be occurring here. But the
14:00really interesting thing they find
14:02is that no difference is
14:03observed in externalizing behavior when
14:05you account for this. So
14:06when you're accounting for the
14:07shared genetic and shared familial
14:09environmental factors,
14:10it seems like this association
14:11with smoking during pregnancy may
14:13not be causal. It may
14:14be a predisposition or epiphenomenon
14:16associated with some of these
14:17confounds.
14:18So because of these reasons,
14:19I think it's it's really
14:20important to try to account
14:22for these these confounds kind
14:23of as as best we
14:24can to, you know, increase
14:25the possibility that the association
14:27we're seeing are predispositional
14:28and or causal, right, or
14:30potentially causal.
14:33So in a series of
14:34studies, we've looked at childhood
14:35outcomes associated with prenatal exposure
14:38in the a, b, c,
14:39d, cohort.
14:40And what you're seeing over
14:42here on the right hand
14:43side
14:44is the, beta estimates, the
14:46standardized effects, and the ninety
14:48five percent confidence intervals.
14:50The red dots and,
14:52intervals are, those who are
14:53exposed both pre and post
14:55knowledge. There are very few
14:56people who are only exposed
14:57post knowledge or pregnancy.
14:58And then in the blue
15:00are the individuals who were
15:01exposed pre knowledge only.
15:03And just to orient you,
15:04essentially, the top part of
15:05the graph for the most
15:06part is externalizing problems. So
15:08when I say externalizing, I
15:09mean things like impulsivity, aggression,
15:11those sorts of things.
15:13And then on the bottom
15:14are more internalizing,
15:15related constructs like anxiety and
15:17depression and the like. And
15:18in the middle, there are
15:19also some total problems, which
15:21is just general psychopathology.
15:23But the main point
15:24is that across the board,
15:26children exposed to prenatal cannabis
15:28have increased rates
15:30of psychopathology
15:31broadly. This is most pronounced
15:32for externalizing. And just to
15:34show you what this looks
15:35like across groups, you can
15:36see a zoomed in plot
15:37of externalizing over here.
15:39And it's actually those exposed
15:41postknowledge of pregnancy in particular
15:43had, much higher rates.
15:45The way that that there
15:47was assessments of frequency and,
15:50of of cannabis being used,
15:51but it wasn't very clear
15:52what those meant, but there
15:53were no differences in reported
15:55frequency of use
15:56among those who used only
15:58preknowledge of pregnancy and those
15:59who used postknowledge of pregnancy.
16:01So it does seem to
16:02be a dose response thing,
16:03although it obviously would be
16:04associated with these these later
16:05stages or more continued use.
16:09The other thing I wanna
16:10point out here is that
16:10we did throw in a
16:13a huge number of potential
16:14confounders, things, you know, like
16:16family history of psychopathology,
16:19other substance use during pregnancy,
16:21prenatal vitamin use, whether the
16:23pregnancy was planned or not,
16:24all of those things. And
16:25then the other important thing
16:26I think we we threw
16:27in here was polygenic risk
16:28for addiction liability in the
16:30child.
16:30So we can take a
16:32an index of general substance
16:33liability from some of our
16:34largest p was that we've
16:35been able to conduct
16:37and estimate the child's genetic
16:38liability to that. And we're
16:40able to account for that
16:41and show that these effects
16:42are present beyond that.
16:44I do wanna be totally
16:46clear. I'm not convinced these
16:47are causal effects by any
16:48stretch of the imagination.
16:50And one thing I wanna
16:51point out is polygenic risk
16:52scores in particular
16:54are
16:54not very they don't have
16:56large effects in predicting their
16:57behavior. Right? So they're you
16:59know, at best, now we're
17:00getting up to ten percent
17:01of the variance in a
17:02phenotype based on this polygenic
17:03scores. Right? So this could
17:05be a little bit of
17:05a straw person argument where
17:07it's just still not the
17:08best index of this, but
17:09the family history stuff is
17:11a pretty robust predictor. So,
17:12you know, I think it
17:13does increase the possibility
17:15that these could potentially be,
17:16you know, causal effects in
17:18some ways. Although, I wanna
17:19be be very clear.
17:21We really need more data
17:22to to to strengthen some
17:23of those those arguments.
17:27One other thing that, we
17:28also found this was associated
17:29with was cannabis use.
17:31So this was a study
17:32where we looked at about
17:33fifty different phenotypes
17:35being associated with early cannabis
17:36use in the ABCD dataset.
17:38So use by age fifteen
17:40of just initiating because that's,
17:42you know, about the amount
17:43of variability we have. And
17:45we found a host of
17:46different, you know, correlates. Looking
17:47at odd ratios here.
17:50And, you know, things like
17:50alcohol use are associated with
17:51an odd ratio of, like,
17:52thirty five. Right? But that's
17:53not particularly interesting because, you
17:55know, kids using one substance
17:56are gonna be likely to
17:57use another.
17:58But the interesting thing is
17:59when you take out those
18:00other, you know, substance use
18:02phenotypes,
18:03the most
18:05predictive variable for predicting early
18:07cannabis use initiation was prenatal
18:08cannabis exposure, and its odd
18:10ratio was approaching three, essentially,
18:12so quite a large effect,
18:13like, a lot of these
18:14other small effects that we're
18:15seeing.
18:16One other thing I just
18:17wanna highlight here that I
18:18think is quite interesting
18:20is,
18:21anhedonia.
18:22Here, if you look at
18:23mid wave, I don't know
18:24if you can see my
18:24cursor. I'm sorry.
18:26But the top blue one
18:28after the golds
18:29is current and past anhedonia.
18:32This is looking at current
18:33past anhedonia baseline predicting future
18:35cannabis use. And I know
18:37there's this kind of stereotypical
18:39anecdotal idea of, you know,
18:41stoners that are becoming, like,
18:42anhedonic like. This suggests at
18:44least that anodontic symptoms are
18:46present kind of before the
18:48initiation of of of a
18:49cannabis use, suggesting that that
18:50could be more of a
18:51predispositional
18:52risk factor for for use,
18:53which I I think is
18:54quite quite interesting.
18:57So when we look at
18:58behavioral conclusions from these data,
19:00we do find that prenatal
19:02cannabis exposure is associated with
19:03a host of adverse child
19:05behavior problems. This is really
19:07across the spectrum. It seems
19:08to be increased genetic liability
19:09for psychopathology.
19:10If we throw instead of,
19:12you know, polygenic risk for
19:13substance involvement,
19:14phenotypes here, if we throw
19:15in polygenic risk for depression,
19:17polygenic risk for ADHD, all
19:18these associations are still there
19:20as well.
19:21So it seems to be
19:22robust for these confounds, but
19:23there obviously, you know, could
19:24be alternative derivations of confounds,
19:26things we're not measuring in
19:27this study that could be
19:28contributing to this association as
19:30well.
19:32One other point I think
19:33is really important to make
19:34with these childhood behavioral outcomes
19:36is that these effects are
19:37generally small. Right? We're talking
19:39about accounting for less than
19:40two percent of variance. So
19:41this is no way to
19:43cannabis exposure would be,
19:45you know, definitively causing this
19:46or would be a robust
19:47huge predictor of this. I
19:49think this really aligns with
19:50much of what we're seeing
19:51in complex behavioral and biological
19:52phenotypes, right,
19:54is that most
19:56potentially ideological things or even
19:57predispositional factors are gonna be
19:59associated with small effects. To
20:01be honest, if I ever
20:01see a large effect, I
20:02start to get worried that
20:03I'm doing kind of circular
20:04research in some way of
20:05looking at, like, you know,
20:06elevated mood and it's associated
20:08with bipolar disorder. You know?
20:10It it gets a little
20:11as soon as we start
20:11to look at these things,
20:12I think these small effects
20:13are what I would expect
20:14from my experience at least,
20:15and a reflection of why
20:17we need some of these
20:17large national studies,
20:19to be able to reliably
20:21identify them.
20:23Alright. So the the next
20:24thing we were kind of
20:25curious about pushing with this
20:26dataset was what are the,
20:27you know, plausible biological mechanisms
20:30that may be occurring.
20:31So here we went to
20:32some of the brain data,
20:33and I I think you
20:34guys are probably, I'm sure,
20:35very well versed with this.
20:36The cannabinoid system
20:39is, you know, involved in
20:40very basic processes, like neurogenesis,
20:42synaptogenesis,
20:43neural migration, axonal, like, elongation.
20:45Right? We know all this
20:46from some really elegant animal
20:48work, right, looking at, like,
20:49the c b one agonists
20:50and and all those data.
20:52And then from some other,
20:53you know, associative studies in
20:54humans, we know it's also
20:55associated with some more complex
20:57biological phenotypes,
20:58things like how the brain
20:59is reacting to threatening information,
21:02learning and memory, cognition, how
21:04fear conditioning is accrued even
21:06through someone like Matt Hill's
21:08really elegant work, looking at
21:10this in some nonhuman animal
21:11models. And,
21:12we also know it's a
21:13search with reproduction and pregnancy
21:14and placental function. Right? Endocannabinoid
21:16levels track with things like
21:17miscarriage and the like, so
21:19they they are also seem
21:20to be very largely contributing,
21:23to to, the intrauterine environment.
21:26And then one other thing
21:27I just wanna highlight is
21:28the association with inflammation, which
21:29I'll show you some data,
21:30are also much less clear
21:32cut than I think the
21:33assumptions often are, right, that
21:35cannabis seems to have an,
21:36you know, suppressant effect or
21:37reduce inflammation.
21:39I'll share some data with
21:40you that where we find,
21:41very strong levels of use
21:42are actually associated with increased
21:44inflammation in this j like
21:45curve, which I I think
21:46is quite quite interesting.
21:48But these are just some
21:49of the data that I
21:50find, you know, very compelling
21:52to implicate the endocannabinoid
21:54system and
21:55THC and cannabis, possibly in
21:56neurodevelopment.
21:58What you're looking at over
21:59on the left is
22:01the data suggesting that WIN,
22:02which is a c b
22:03one agonist you could think
22:04of it as,
22:06leads to axonal repulsion.
22:08So, you know, these these
22:09growth currents will will develop
22:10out and with BDNF, for
22:11example, you'll see it growing
22:13towards where the BDNF is
22:14no is, placed as indicated
22:16by the the white, cursor
22:18there. But when you add
22:19WIND, for example, which is
22:21you can think of as
22:22being functionally similar to THC
22:24in many ways, but acting
22:25on c v one and
22:26agonizing it, That leads to
22:28growth cone repulsion where the
22:30neurons will actually grow away
22:31from that area. So you
22:33can see how this could
22:33lead to things and, you
22:34know, arguably even things like
22:36psycho
22:37architectural differences associated with schizophrenia
22:39risk and all of those
22:40sorts of things that that
22:41that,
22:42you know, you could see
22:43the endocannabinoid system
22:45impacting.
22:46And then you also see
22:46in adults, at least, you
22:47know, reductions in new synapse
22:49formation and including in in
22:50rhesus macaque models. And then
22:52there is some evidence of
22:53prenatal cannabis exposure increasing GABA
22:56interneurons in the hippocampus and
22:57some some nonhuman animal models
22:59as well. So I think
23:00it's very plausible,
23:01obviously,
23:02that the endocannabinoid system and
23:04changing this at these developmental
23:05time periods could have, you
23:07know, potential consequences on growth
23:09and development.
23:11So to do this, we
23:12looked at some of the
23:12neuroimaging data that's involved in
23:14the ABCD study. We had
23:15two time points. So we
23:17had one time point at
23:18baseline when the children were
23:19on average nine to ten,
23:20and then another one when
23:21they were, on average eleven
23:23to twelve.
23:25There is partial data release
23:26to the other one, but
23:27we find there being differences
23:29in kind of exposure and
23:30risk risk, variables in that,
23:32and that it seems like
23:32some of the kids that
23:33came into the study later
23:34were more enriched for for
23:36problems. So because of that,
23:37we restricted our data to
23:38these two year data, but
23:39we're expecting a new data
23:40release really soon.
23:42In this study, we didn't
23:43feel like we had great
23:44priors for what we would
23:46expect. So we took a
23:47kind of all encompassing approach
23:49and just looked at all
23:50different indices of brain structure
23:52and resting stage functional connectivity
23:54here. We did look at
23:55task related data because of
23:56some problems with the reliability
23:58associated with it. Although, I'm
23:59I must say, I think
24:00there are some exciting developments
24:01there that David Behringer is
24:03doing and leading actually, looking
24:04at neural signatures, and we're
24:06increasing the reliability of some
24:07of those. I'm hopeful to
24:08look at those in the
24:08future as well.
24:10But we essentially took an
24:11all comers approach and just
24:12corrected for multiple testing correction
24:14across all of these different
24:15tests that were we were
24:16running. So there could be
24:17some false negatives in here
24:18without a doubt, but we
24:20think we're pretty well protecting
24:21against false positives.
24:23And then we tried to
24:24see if any of these,
24:26neural differences that are associated
24:28with prenatal cannabis exposure may
24:30be associated with future child
24:31behavior. So we do have
24:33data time points a year
24:34later on the behavior of
24:36these kids. So we could
24:36look at longitudinal mediation. So
24:38it's prenatal exposure associated with
24:39brain differences at baseline in
24:41two year and the change
24:42in those, and then is
24:43that associated with future behavior.
24:46And we do find several
24:48interesting associations. So what you're
24:50looking at at the dotted
24:51line
24:52is,
24:53FDR correction across all of
24:55the tests run.
24:56And then underneath this, anything
24:58that is bolded,
25:00in a circle is representing,
25:02FDR correction within that modality.
25:05Because one can argue these
25:06are different questions. Right? White
25:08matter looking at white matter
25:09differences and DTI or something
25:10is very different than looking
25:11at gray matter differences or
25:13resting state functional connectivity.
25:15So,
25:16and look at you could
25:17consider these lower threshold of
25:19significance, but also showing differences.
25:22There are a couple of
25:23findings I wanna just highlight
25:24here, and then we'll focus
25:25on a couple of them
25:26that are showing some longitudinal
25:27mediation.
25:28One is there is this
25:29interesting difference associated with the
25:31resting state functional connectivity. So
25:33this is looking at how
25:34functional activity measured with fMRI
25:37in one area is correlating
25:38with other other areas of
25:39the brain. So you're not
25:41looking at response to anything.
25:42You're just looking at how
25:43these areas are tracking together
25:44in their activity.
25:46And the interesting thing we
25:47found here was that the
25:48auditory network, this broader,
25:51network associated with auditory processing
25:52was showing increased connectivity among
25:54those with prenatal cannabis exposure
25:56with the left, putamen.
25:58I find this this is
25:59all, you know, kind of
26:01I I find this somewhat
26:02interesting just because of
26:04some tangential evidence in schizophrenia,
26:07linking these, for example, to
26:09hallucinations and and schizophrenia
26:10risk in some ways. So
26:11it I think there could
26:12be something there, but we're
26:13not seeing that, you know,
26:14by any stretch of the
26:15imagination here, but I thought
26:17it was an interesting finding.
26:18And then the other big
26:19thing I wanted to point
26:20out is that DTI measures
26:23or diffusion tensor weighted imaging
26:24measures,
26:25this is generally track tracking
26:27white matter, integrity,
26:29but but then it can
26:30also track some putative indices
26:32of neuroinflammation,
26:33particular in gray matter. And
26:35that's where we're finding the
26:36enrichment of these findings.
26:38So I do think this
26:39could nicely align with some
26:40of these, you know, nonhuman
26:41animal models and, you know,
26:43emphasizing axonal guidance systems and
26:44the like,
26:46and could potentially align with
26:47that. So all the DTI
26:49and RSI measures are these
26:51diffusion weighted imaging measures that
26:52are gonna be reflecting how
26:54the white matter is structured,
26:55and then also some indices
26:57of, you know, could be
26:58things like neurodensities in some
27:00gray matter areas as well
27:01as white matter density there,
27:03but also some, inflammation related
27:05cellularity, particularly in the RSI
27:06measures. So I wanna walk
27:08through a couple of these
27:08findings in a little, greater
27:10detail.
27:12So the next thing we
27:13did was we looked at
27:14all of these correlates that
27:15were significantly associated with this
27:16after, you know, multiple testing
27:18correction. We said, are they
27:19associated with future behavior in
27:21these kids? And we do
27:23find that they are. We've
27:25there are about eight of
27:25them that are, but there
27:27were four of them that
27:29are actually mediating future behavior.
27:31And it's in the direction
27:32that you'd expect.
27:33So, you know, any prenatal
27:35cannabis exposure is associated with
27:37these differences in restricted total
27:38diffusion as well as transverse
27:40diffusivity in the pars granularis.
27:42And these are in turn
27:44associated with future,
27:46externalizing behavior. The biggest hits
27:48were for ADHD related problems
27:50in these kids.
27:51So I wanna just walk
27:52through some of our interpretations
27:53of what this could mean.
27:55So
27:56isotropic diffusion or nondirecturally dependent
27:59diffusion,
28:00has been postulated to be
28:02a potential marker of neuroinflammation.
28:04And in particular, when there
28:05are increases in mean and
28:07transverse
28:08cortical diffusivity, so in these
28:10cortical regions,
28:11and restrict reduced restrictions
28:14in directional and total diffusion
28:16of white matter bundles,
28:18that is typically thought
28:20to be a sign of
28:21of neuroinflammation.
28:23So what we actually find
28:24is the opposite of this.
28:25We find reduced isotropic diffusion,
28:28less diffusion in these these
28:29broad areas.
28:30And so that could putatively
28:33reflect less neuroinflammation
28:34associated with prenatal cannabis exposure.
28:37And, you know, some of
28:38these pruning
28:39mechanisms that occur to help,
28:40you know,
28:42take away,
28:43you know, ipsilateral connections,
28:45in retinal development and all
28:46those have been well identified
28:48and characterized,
28:49with regard to risk in
28:50schizophrenia.
28:51So it's possible that, you
28:52know, too much or too
28:53little here could have have
28:54problems for, you know, early
28:56neurodevelopment.
28:57And then the other possibility
28:59that that we see, you
29:00know, this collection of findings
29:01could indicate
29:02is that when there's
29:04this decreases or reduced
29:06cortical transfused diffused diffused that
29:08you said this would be
29:09along these lines
29:11and,
29:13increases in restricted directional diffusion,
29:16so on these these white
29:17matter tracts that are developing
29:18during adolescence,
29:20there is some evidence that
29:21that pattern generally will be
29:22occurring as adolescence progresses progresses.
29:26So another possibility of these
29:28findings collectively, we think, is
29:30that this could reflect accelerated
29:31neurodevelopment
29:32among these kids that have
29:33prenatal cannabis exposure. And there's
29:36some anecdotal, you know, data
29:37that could be consistent with
29:38that. You know, sometimes you'll
29:40see that kids with older
29:41friends are more likely to
29:42engage in cannabis use perhaps
29:44due to a greater exposure.
29:45Sometimes we even see greater,
29:47language abilities in kids who
29:49are engaging in substance use
29:51earlier.
29:52So it it could be
29:53associated with those things, but,
29:54you know, we're just starting
29:55to explore these relationships. So
29:56it I'd welcome also other
29:58ideas that that people may
29:59have.
30:01Alright. So before moving on
30:03to some of our cuddle
30:03work, I I do wanna
30:04just highlight some limitations and
30:06challenges to this work overhaul.
30:08One is in the ABCD
30:09study, we're relying on retrospective
30:11report of use during pregnancy
30:12only.
30:13You know, as I've shared
30:14with you from some of
30:15the New York Times stuff
30:16and and the like, it
30:16does seem like people are
30:17not unwilling to necessarily admit
30:20this, and we're we're seeing
30:20that in a lot of
30:21our other data and that
30:22our self report data really
30:24tend to align with biological
30:25data of, you know, urine
30:27drug testing,
30:29or even umbilical cord tissue,
30:31as well, which is probably
30:32the best way to characterize
30:34this in reality.
30:36And there are some data
30:37that they've
30:38of pregnancy studies where they've
30:40assessed them during pregnancy and
30:41asked for what medications they're
30:43using
30:44and substances they use, and
30:45then they asked them, like,
30:46five and ten years later.
30:48And they actually do find
30:49those to be relatively correlated.
30:51So I think there's some
30:52evidence to suggest that perhaps
30:53during pregnancy, we're more attuned
30:55to what is being used
30:56and all of those sorts
30:57of things in this retrospective
30:58report. Maybe accurate, but I
30:59I do think it's a
31:00problem. Right? And I don't
31:01think there's an easy solution
31:02to this, because some of
31:04the testing differences associated with
31:05urine and how long some
31:06of these, you know, substrates
31:07stay and how it's associated
31:08with BMI and adiposity and
31:09all of those things. Right?
31:11So I think the biological
31:12data is gonna be great
31:13for capturing regular chronic use,
31:15but, you know, in a
31:16interspersed use, it may not
31:17be the gold standard necessarily
31:19either.
31:20Another thing is there's a
31:21large temporal gap here. Right?
31:23There's ten years between the
31:24behavior we're looking at and
31:25the brain differences we're looking
31:27at, and it's likely that,
31:28you know, you know, have
31:29this exposure associated with a
31:30lot of those things
31:31that, you know, could be
31:32making this seem like a
31:33potential causal influence that isn't
31:35there. We have done some
31:36subsequent analysis where we've covariate
31:37for parenting styles and those
31:39sorts of things, and we
31:40still find the relationship there.
31:41But, obviously, what we really
31:43need to do is follow
31:43them during this period of
31:44time and and, you know,
31:46try to, you know, better
31:47account for some of these
31:48these temporal plausibility
31:49plausibility.
31:50I wanna be, you know,
31:52very clear. While these data
31:54by accounting for these confounders
31:55that increase the possibility there
31:57could be causal effects,
31:58they're we're not demonstrating that
32:00at all. These are just
32:01associations.
32:02And then lastly, in ABCD,
32:04despite being a relatively large
32:05sample, we still do have
32:06relatively small numbers of, children
32:08who are exposed. And, you
32:10know, these system and more
32:11molecular level, mechanisms remain really
32:13poorly understood.
32:15So in conclusion, there is
32:16a small increase in risk
32:17for adverse behavioral outcomes in
32:19childhood,
32:20Whether this represents causal shared
32:22predisposition and or epiphenomenon
32:24in some way is really
32:25unclear.
32:26The neural correlates, we are
32:27seeing that structural and functional
32:29connectivity metrics do partially longitudinally
32:31mediate this as well.
32:33So, you know, this could
32:34reflect accelerated neurodevelopmental
32:37development. It could reflect reduced
32:39inflammation and neural inflammation in
32:41particular during early,
32:43child development.
32:45But, you know, the what's
32:46causing that is unclear, but
32:47I do find it quite
32:49interesting that we are seeing
32:50these these mediational pathways where
32:51it's being associated with both
32:52exposure and,
32:54the the future behavior in
32:56these kids.
32:58And there are some other
32:59data that do kind of
33:00align with this idea of
33:02inflammation in some ways.
33:04What you're looking at here
33:05are data from a TPSO
33:06PET ligand study.
33:07So TPSO
33:09is a a mitochondrial gene
33:10that's expressed in
33:13and it is expressed alongside
33:14microglial activity. So microglia are
33:16the, you know, immune centers
33:18of the of the of
33:19the brain and among many
33:20other functions.
33:21But this is looking at,
33:23this putative index of neuroinflammation
33:25and a small sample of
33:26around twenty five ish or
33:27so,
33:28regular cannabis users and controls.
33:30And across brain regions, what
33:32you can essentially see is
33:33in those who are
33:34long term users on the
33:36left, as well as those
33:37with a quantified problem of
33:39use with cannabis use disorder
33:40on the right is are
33:41these indexes of higher inflammation.
33:44We are finding indexes of
33:46potential lower inflammation, you know,
33:48of in the in the
33:49the prenatal data. But there
33:51are some other data that
33:52are could potentially be consistent
33:54with this. So these are
33:55some preliminary data where we've
33:56looked at, relationships between cannabis
33:59use and inflammation in adults.
34:01So the data on the
34:02left are from the UK
34:03Biobank.
34:04So this is data in
34:06about a a hundred and
34:06fifty thousand or so older
34:08adults, both middle aged and
34:09older adults that would live
34:11in the UK.
34:12And you see this and
34:13all these line patterns represent
34:15significant differences, essentially.
34:17But there's, kind of a
34:18what I like to think
34:19of as a kind of
34:19a j shaped pattern
34:21whereby,
34:23never use, if you're looking
34:24at CRP as a one
34:25index of neuroinflammation as a
34:27really stable, nice, you know,
34:28easy easy measure.
34:30People who are using cannabis
34:31less frequently
34:32actually have relatively lower levels
34:35of peripheral inflammation than those
34:36who are not using it,
34:38or who don't report using
34:40it regularly and never used
34:41it. But those who are
34:42using daily and very regularly
34:44and, again, there's not relatively
34:46small sample size in in
34:47this dataset. But there, we're
34:48actually seeing the inverse pattern
34:49where we're actually seeing increased
34:51levels of neuroinflammation.
34:53And we do see some
34:54other data that are consistent
34:55with this. So the data
34:57on the right is a
34:58genomic structural equation model looking
35:00at GWAS data. So these
35:02are separate GWASs of cannabis
35:03use disorders, separate GWAS of
35:05c reactive protein, of BMI
35:07cigarettes per day, and educational
35:08attainment.
35:09And what we essentially see
35:10is that problematic cannabis use
35:12or cannabis use disorder
35:14has a
35:15positive genetic correlation with CRP
35:18even when we had taken
35:19to account these other potentially
35:20confounding factors of BMI, cigarettes
35:23per day, and educational attainment
35:24here. But the interesting thing
35:26is if we sub out
35:26cannabis use disorder and instead
35:28put out ever use of
35:30just ever having used cannabis,
35:31we actually find the directionality
35:33flip consistent with this UK
35:35Biobank data
35:36where, ever cannabis use is
35:38associated genetically correlated
35:40with lower,
35:42CRP levels.
35:43So I think this is
35:44interesting, because it it does
35:45align with this idea that
35:47perhaps the
35:48links between inflammation and cannabis
35:50use are much more complicated
35:51than we may be led
35:52to believe by some of
35:53the the the initial studies.
35:55And And it really suggests
35:56that problematic heavy cannabis use
35:58may be associated,
36:00you know, with with these
36:01elevations and inflammation. And with
36:03these data, we have tried
36:04to push it. Like, we
36:04tried to, you know
36:06perhaps this is a you
36:07know, the regular users are
36:08using it for pain, which
36:09obviously is associated with inflammation
36:11or other medical problems. And
36:13even when we covary for
36:14that, we still see this
36:15the these these, associations.
36:17So I think it's quite
36:18interesting, but we really need
36:19to push these data further.
36:20I very much see them
36:21as preliminary at at this
36:22point still.
36:25So lastly, I wanna talk
36:26about one of the studies
36:27that I've, been doing with
36:29Cynthia Rogers and Arpana Agrawal
36:30here to try to address
36:32some of these limitations
36:33of the,
36:35of the ABCD dataset.
36:36So here with our in
36:38collaboration with our OBGYN clinic
36:40here, we're recruiting women who
36:41are using cannabis and those
36:42who are not.
36:44And one critical thing about
36:45this study that I think
36:45is nice, we're trying to
36:47account for these confounds,
36:48is even though
36:50the individuals who are not
36:51using cannabis during their pregnancy,
36:52we require them to have
36:54endorsed lifetime use of cannabis.
36:56So, you know, hopefully that
36:57will account for some of
36:58the confounds associated with ever
37:00exposure.
37:01And we are recruiting them
37:02all from the same clinic.
37:04And one of the other
37:05interesting thing is there are
37:06some racial differences epidemiologically
37:08associated with cannabis use, where
37:10in black samples, you'll tend
37:12to see cannabis use used
37:14in isolation of other other
37:15substances without alcohol and the
37:17like. So partly because of
37:18our population here in Saint
37:20Louis, it's a predominantly black
37:21population, but we're able to
37:22control and recruit a sample
37:24that's not using other substances,
37:26which is one of our
37:27eligibility requirements.
37:28With that said, we are
37:30getting a fair amount of
37:31cotinine positivity. And when we
37:32look at the self report
37:33data,
37:34they are reporting using blunts,
37:36and also there's a common
37:37thing too here of wrapping,
37:39Tibet the joints as well
37:41in cannabis sleeves. So there's
37:42kind of specific thing in
37:43in addition to blunt usage.
37:45So we think it's occurring
37:46from that. They'll report they're
37:47not using cotinine or tobacco
37:49or anything, but they then
37:50they'll report that they're using
37:51it in this way.
37:53A couple things I really
37:54want do wanna point out
37:54with the sample, though, is
37:55it's matched in
37:57non,
37:58you know, you know, cannabis
37:59and cannabis exposure groups, but
38:00it's a very under resourced
38:01population.
38:03You know, about fifteen percent
38:04did not complete high school.
38:06Almost fifty percent are living
38:08beyond below the federal poverty
38:09line, which is quite quite
38:11low. And some of those
38:12factors are associated with testing
38:13positive for coat name, for
38:14example, and but not reporting,
38:16tobacco use, which,
38:18we think could be playing
38:19a role.
38:20But there are some interesting,
38:22patterns we are seeing. First
38:24off, we did do urine
38:25drug screens in this. So
38:26this is not the gold
38:27standard. We wish we could
38:28have done placental cord tissue
38:29and the like. We're looking
38:30to do that in the
38:30future in this sample because
38:32we did collect those data.
38:33But these are just the
38:34urine dipsticks, right, that don't
38:35have the that great temporal,
38:37resolution.
38:38But we do find them
38:39to be concordant in most
38:40cases,
38:42for so those who are
38:43reporting use are also testing
38:44positive. Those who are not
38:45reporting use are not testing
38:47positive.
38:47We do have a substantial
38:49portion not a substantial portion,
38:51but around it's approaching twenty
38:52percent or so
38:54of individuals who don't report
38:56use but are testing positive.
38:58And in some of those
38:59cases, we think there are
39:00some possible explanations. In the
39:01literature, this is all typically
39:03reported to be deception,
39:04from individuals or that they're
39:05lying about their use during
39:06pregnancy, and I think there's
39:07evidence that suggests this could
39:08be other things.
39:10For example, there's some data
39:11that to suggest that secondhand
39:13exposure in confined rooms regularly
39:15can lead to testing positive.
39:16And a good portion of
39:17these women who were testing
39:19positive for this
39:20but reporting no use,
39:23reported
39:24living in a home where
39:25there was a regular smoker
39:27or, living
39:28or you're traveling in a
39:29car regularly,
39:30with somebody who is using.
39:32The other thing I wanna
39:32note about our sample is
39:33we did assess all kinds
39:34of different modes of use
39:35in this.
39:36In this sample, it's almost
39:38it's I wanna say it's,
39:39like, ninety five percent smoking
39:40or using Bongs.
39:41So it's very traditional measures.
39:43This is not, like, a,
39:44you know, a higher SES
39:45sample that's using gummies and
39:47all of those sorts of
39:48things. It's really a a
39:49smoking sample.
39:51So, yeah, we've we've tracked
39:52birth outcomes in them. We
39:54are seeing the reduced reductions
39:55in birth weight,
39:57but I remain unconvinced that
39:58this is independent of coat
39:59name. One of the difficulties
40:01is our heavier cannabis users
40:02are also the ones testing
40:03positive for coat name through
40:05blunt usage, presumably,
40:07but there could be other
40:07factors as well. And this
40:09this does seem to be
40:10there, but it's certainly less
40:11strong. So at least some
40:12of it seems to be
40:13potentially attributable to comorbid,
40:15cotinine positivity.
40:17And then the other thing
40:18that you're looking at down
40:20here on the bottom graph,
40:21the IBSI,
40:22this is an infant temperament
40:23test that is looking at
40:25externalizing
40:25behavior.
40:26And at six months of
40:28age actually, excuse me. This
40:29is twelve months of age.
40:31And this is a subsample,
40:32so we're still getting there.
40:32So this is very plain,
40:33hey. It's a subsample of
40:34the data.
40:35But we are seeing increased
40:37externalizing like behavior, then as
40:39well, associated with prenatal, cannabis
40:41exposure.
40:43And then lastly, one of
40:44the other things we've been
40:45doing is trying to attack,
40:46you know, why are people
40:48using cannabis during pregnancy? And
40:49this this is the full
40:50data set. Right? Because we've
40:51had everybody give birth now
40:53essentially.
40:54But we asked, women to
40:55report on why they were
40:56using cannabis or what their
40:57motives were for using cannabis.
40:59And people could endorse multiple
41:01of these, so they're not
41:02mutually exclusive. Right? But we
41:03sound you know, that most
41:05women were reporting using it
41:06for nausea or vomiting in
41:08the first trimester.
41:09But the other thing that
41:10came out was,
41:12one, regulating things like appetite
41:13or sleep, but also a
41:15lot of psychopathology. So you
41:16see people reporting to use
41:17it for anxiety, stress, depression,
41:19PTSD,
41:20energy, pain management, and the
41:22like.
41:23And these differences in reports
41:25are also associated with continued
41:27use. So these are the
41:28motives that they reported in
41:29the first trimester.
41:31And if we look at
41:31that and say, well, if
41:33these are your motives of
41:33use, how likely is it
41:35that you'll continue to use
41:36into into the pregnancy?
41:38And,
41:39we do find that essentially
41:41all of these more psychopathology
41:43related factors that people are
41:44reporting
41:45use, of their reason for
41:46using,
41:47that is predictive of a
41:49two to threefold odds ratio
41:50of continued use into the
41:52second and third trimesters.
41:53But the other interesting thing
41:54was that those who are
41:55reporting use for nausea over
41:57the the far right,
41:58that was not associated with
41:59a significant increase or
42:01a significant likelihood of continuing
42:03use through the pregnancy.
42:05And some of that may
42:05be, you know, reflective of
42:07when nausea is most, you
42:08know, prominent in in pregnancy
42:09during the the first trimester,
42:11for example, whereas other things
42:12like pain and some of
42:13these psychopathology,
42:14problems are are more consistent.
42:17And then lastly, on the
42:18the bottom graphs
42:19are just showing,
42:21depression and stress,
42:23trajectories through pregnancy across the
42:25the three different trimesters.
42:27The individuals with the the
42:29gold are the gold colored
42:31dots are those who used
42:33a prenatal
42:34cannabis during their pregnancy
42:36and reported a motive of
42:38using it for mental health.
42:40The individuals in blue are
42:41reflecting of people who used
42:43cannabis,
42:44the light blue, who used
42:46cannabis during their pregnancy,
42:48but reported other motives for
42:49using it. So not things
42:51like struggling with depression, for
42:53example.
42:54And then the other the
42:55darker blue line are those
42:56who didn't use cannabis.
42:58So one is the people
42:59that are report using cannabis
43:00and report using it for
43:01mental health are reporting more
43:02stress, more depression.
43:04But we're not seeing changes
43:05across pregnancy. There's actually, if
43:07you look at the slopes,
43:07so if you do the
43:08longitudinal data analysis kind of
43:10across trimester, we find no
43:11difference in change. It's just
43:12the intercept. They're they're I
43:14don't think it's particularly surprising.
43:16But, you know, and we
43:17did do some follow-up analysis
43:18that aren't reported here where
43:19we took,
43:21other people who were not
43:23using cannabis, for example, that
43:24had high
43:25depression scores at baseline, and
43:26we find their trajectory to
43:28largely mirror, these individuals. So,
43:30you know, whether it's helping
43:31or maybe we can't tell
43:32from these data, but it's
43:33not really associated with the
43:34change, over over time.
43:37Cuddle limitations. There are a
43:39host of limitations here. Right?
43:40The generalizability
43:41is limited. This is a
43:42really undersourced
43:43and predominantly black sample. It's
43:45about ninety percent, black.
43:48While it's somewhat large for
43:49a, you know, a study
43:50of prenatal, and we're collecting
43:51neonatal neuroimaging data on them
43:53in their first month of
43:53life,
43:55and then we're following them
43:56and we're just getting a
43:57renewal, we think, to follow
43:58the month until age five
43:59and six. It's still a
44:00relatively small sample, so we
44:02might be underpowered.
44:04And then our phenotyping and
44:05our frequency of assessment was
44:06constrained.
44:07So we have this neuroimaging
44:08data from zero to one
44:09month,
44:10and then we have behavioral
44:11assessments at six, twelve, and
44:13eighteen months, and then our
44:14renewal grants, would be getting
44:16imaging at five and six
44:17and behavior there. So there's,
44:18you know,
44:20where there's a lot of
44:20development going on, we're gonna
44:22have some gaps in our
44:22in our study here.
44:25So with that, I do
44:26wanna just briefly mention
44:27the HPCD study. So this
44:29is another national,
44:30study that was kind of
44:31inspired by ABCD. It is
44:33across twenty five sites with
44:34one of the sites being
44:35us here at WashU
44:37that are trying to recruit
44:38around seven thousand seventy two
44:39hundred pregnant individuals.
44:41Twenty five percent of them
44:42with substance use, and this
44:44would be substance use qualified
44:45as opioids, tobacco, alcohol, or
44:47cannabis.
44:48And then following their offspring
44:50up until age ten and
44:52hopefully beyond that even.
44:54And within this, we're doing
44:55a lot of intensive sampling
44:56in early life. So right
44:57here, you can see our
44:58first visit comes with the
44:59typically the second trimester,
45:01and then visit two has
45:02neuroimaging and a bunch of,
45:04you know, physiological measures and
45:06biosamples and the like at,
45:08zero to two as a
45:09neonate at zero to one
45:10month. And then we follow
45:11them at potentially through infancy
45:13and then start to spread
45:14these assessments out every year,
45:15year and a half. We're
45:16working on the renewal grant
45:17for it right now, so
45:18the the exact frequency is
45:20a little unclear,
45:21still as it goes beyond
45:23visit six.
45:24But we are getting biological
45:25data from these. Eleven sites,
45:27including ours, are also collecting
45:28placental samples in core tissue,
45:30in cord blood in the
45:31leg. So I hope that
45:32that that will be really
45:33useful for better characterizing the
45:35intrauterine environment.
45:37But so I'm really excited
45:38about looking at these data.
45:39I think these data may
45:40be more, you know, limited
45:41for looking at cannabis specifically,
45:42although that is the most
45:43frequently used, you know, substance
45:45in this this sample out
45:46you know, in isolation too.
45:48So I'm hopeful that it
45:49may be useful there. But
45:50there's a lot of comorbid
45:51use.
45:52And then just one other
45:53thing for the sample is
45:54that,
45:55we're really pushing our opioid
45:56recruitment right now. We're trying
45:57to get twelve and a
45:58half percent of the sample
45:59to have opioid recruitment.
46:01So there is a a
46:02lot of the additional cases
46:04that are gonna be added,
46:04I think, are gonna have,
46:06some some comorbid,
46:07treat, uses
46:09as well. But these data
46:10will be made available in
46:11the near future. We're thinking
46:13January, but all of the
46:14the hiccups that NIH has
46:15placed in this with some
46:16of the new administration and
46:17this compliance and all of
46:18that is slowing this down.
46:19So I'm hopeful that at
46:20this summer, at least, hopefully,
46:22these data will be released.
46:23And, again, these are are
46:24available to anybody. So, would
46:26strongly encourage people to to
46:27work with these data. As
46:29far as future directions,
46:31with Emma Johnson, who's
46:33a assistant professor here, we
46:35have collected placental samples in
46:37our CUDL dataset,
46:39and we have generated a
46:41methylation data from them. So
46:42we're very curious about looking
46:43at methylation
46:44and, you know, looking at
46:45pathways that methylation
46:47may be implicated in and
46:48seeing if some of those,
46:49you know, align with some
46:50of what Yasmin Herb's group
46:51has has found and
46:53and the like. And then
46:54also what I'm really curious
46:55about doing there is seeing
46:56if there's enrichment
46:58in genes. So, like, is
46:59there enrichment for methylation signatures
47:01in genes that are then
47:02associated with genome wide association
47:04studies of psychopathology, for example?
47:05So are we seeing that
47:06the the genes are differentially
47:08impacted?
47:09We are working with some
47:10neural novel measures of neuroinflammation
47:12using some diffusion weighted data.
47:14Yong Wang here has developed
47:15what's known as DBSI or
47:17diffusion basis spectrum imaging, and
47:19has some really nice validation
47:20data showing that these indices
47:22seem to correlate really well
47:23with, you know,
47:25immunohistochemistry
47:26stains,
47:27measures of cellularity,
47:28and some rodent models and
47:29the like. So we're starting
47:30to work with some of
47:31those data, and then we
47:32are working on some,
47:34inflammation data in kids. And
47:36we're also working on some
47:37stimulated. So not just looking
47:38at basal circulating levels, but
47:39if you hit it with
47:40a pathogen like LPS,
47:42you know, what is the
47:42capacity it has to respond
47:44to this?
47:45And then, you know, obviously,
47:46we're very interested in winding
47:47the clock back and looking
47:48at these neonatal and infant,
47:50and young child MRI. And
47:51then outside of that, you
47:52know, there is a sort
47:53of mating.
47:54The, you know, individuals who
47:56are pregnant who are using
47:57cannabis
47:58are likely with in relationships
47:59with are gonna be more
48:00likely to be in relationships
48:01with people with men who
48:03are using cannabis,
48:05perhaps while, sperm cells were
48:06being generated.
48:07And there is some, you
48:08know, interesting data suggesting some
48:10epigenetic signatures in sperm cells
48:12as well. So we're just
48:13starting a preliminary data data
48:15collection here where we're
48:17collecting, seminal fluid samples from,
48:19kind of heavy cannabis using
48:21men as well as controls.
48:23It was fun getting that
48:23to our department and getting,
48:25I can't remember what the
48:26name of it is, but
48:27some of the pornography website
48:28thing that we had to
48:29get to help men do
48:30this and getting funding from
48:31that from our university is
48:32quite quite quite the fun,
48:33but we're we're getting there.
48:35But we're really interested in
48:36looking at some of these,
48:37you know, paternal germline cells
48:38too,
48:39and, plus, why I think
48:40genetic signatures. So with that
48:41said, I wanna thank everybody
48:43in the lab, acknowledge them,
48:44and acknowledge our sources, and
48:45thank you all for listening.
48:46I'd be happy to to
48:47chat further.
48:52Thanks, Ryan. That was,
48:54that was fascinating.
48:56I'm inviting people to ask
48:57questions.
48:58You know, if you can
48:59either put questions in the
49:00chat
49:01or you can unmute yourself
49:02and identify yourself and ask
49:04the question.
49:09So May I ask you
49:10the first question? Go ahead,
49:11Annie. Anita.
49:13Hello.
49:14I'm Anita. I'm an assistant
49:16professor of psychiatry here working
49:17with Cyril.
49:19So if I remember correctly,
49:22the
49:22individuals who who were using
49:25cannabis during pregnancy
49:26also had lower,
49:28like, financial,
49:29social economic. Absolutely.
49:32We're very able to maybe
49:33compare
49:34at least a relative risk
49:35of, socioeconomic,
49:39maybe stress on kinda, like,
49:41and compare it and, like,
49:43order
49:44contribution of that in the
49:45Yeah. Yeah. So so we
49:46did so one, our analysis
49:48did have SES as a
49:48covariate along a host of
49:50host of other ones. The
49:51other thing we did was
49:52we did supplemental analysis of
49:54propensity score weighting.
49:55And we did propensity score
49:57matching as well, which is
49:58where essentially you can think
49:58of it. A simplest way
49:59of thinking of it is,
50:00like, you're generating, like, a
50:02factor of all your covariates,
50:03so you are not loading
50:04up your model with so
50:05many covariates. It can be
50:06a nice way that you
50:07don't hurt your statistical models
50:08as badly with, you know,
50:09having so many covariates, but
50:11I also don't like it
50:11as much because then you
50:12can't see the associations of
50:13those covariates with the outcomes,
50:15as as clearly.
50:17So we we do see
50:18those associations there. We do
50:19see it if we do
50:20propensity score matching where we
50:21you know, you can think
50:22of it as a one
50:23to one matching of, you
50:24know, across these, potentially confounding
50:26variables in the exposed versus
50:28the nonexposed,
50:29or if we do this
50:30propensity score weighty, which is,
50:32you know, essentially, you could
50:33think of it as getting,
50:33like, a factor score for
50:34everybody all across these covariates.
50:37So we we do see
50:38that they're independent
50:39of of of that, but
50:40it you know,
50:41it you know, I'm not
50:42convinced that it's not contributing.
50:45Is there any, like because
50:46there are some evidence that
50:48the synergism
50:49between the stress and cannabis
50:51use may kinda likely to
50:53more psychosis. Is there any
50:55do you see any evidence
50:56of that? Yeah. I think
50:57that's a fascinating question. I
50:58wish we could look at
50:59it. I'm somebody who's very
51:00skeptical of interactions with with,
51:02small samples coming from a
51:04g by e world,
51:05and a lot of, you
51:06know, initial false positives. So
51:07I think we have to
51:08be really well powered because,
51:09you know, soon as we
51:10look at, you know, dividing
51:11things up by high stress,
51:11low stress, prenatal cancer exposure,
51:13no prenatal cancer exposure ourselves
51:14are gonna get really, really
51:15small.
51:16And then I'm also a
51:17firm believer that to fully
51:18account for interaction compounds, you
51:20have to do the covariant
51:21of no interest by predictor
51:22of interest interactions. So you'd
51:24be wanting to, you know,
51:25load your models up with
51:26sex by, you know, cannabis
51:27exposure, sex by stress exposure,
51:29you know, all of those
51:30things. So I don't think
51:31we're there with a well
51:32powered sample enough, but I'm
51:33hopeful that HPCD actually may
51:35be able to get us
51:36there where we might be
51:37able to more reasonably look
51:38at some of those things.
51:39But I think you're like
51:40you're probably entirely correct. These
51:43all these associates are much
51:44more complex than we can
51:45look at,
51:46but we feel like we're
51:47able to look at more
51:47of these main effects now.
51:49But I think there's likely
51:50all kinds of, you know,
51:52you know, gene by environment,
51:53environment environment
51:55interactions, and, you know, gene
51:57environment correlation, I think, is
51:58an underappreciated thing as well,
52:00that is often not well
52:01accounted for.
52:03And we are starting to
52:04work with some of these
52:04data too with we can
52:06look at a prenatal cannabis
52:07exposure as well. But we're
52:09starting to look at some
52:10discordancies among the twin pairs
52:11and sibling pairs and a,
52:13b, c, d as well
52:14to try to, you know,
52:15better account for those sorts
52:17of confounds. Brian, I have
52:18a, great great talk. I
52:20have a fairly basic question.
52:22And could you comment on,
52:25how one might tease out
52:27the effects of
52:28the
52:30constituents of can of cannabis
52:33from,
52:34from just the pyrolysis
52:36of herbal material,
52:38and what contribution do these
52:39play in whatever negative outcomes
52:42that we're seeing? Yeah. I
52:43think this is a great
52:44question. I think that, you
52:45know, the different constituents of
52:46cannabis or all of the
52:47additives, especially with, you know,
52:49people that are using things
52:49in different things, is this
52:51from the cannabis, or is
52:52it from these other things?
52:53I think we totally don't
52:54know.
52:55There are you know, I
52:56mean, there there are being
52:57increasingly some commercial available things.
52:59And given the legal changes,
53:00I'm hopeful that we'll be
53:01able to better, you know,
53:02do this. But there are
53:03some studies where they're having
53:04people bring in their samples
53:06and quantifying it for these
53:07these different things. So I
53:08think that's really the best
53:09way forward. And some areas
53:10in your some studies in
53:11Europe have been done with
53:13with this to some degree
53:13as well.
53:15In our sample, most of
53:16our sample is still getting
53:17at more on the black
53:18market. So, you know, if
53:19we ask them to report
53:20on potency or any of
53:21those sorts of things as
53:22as well,
53:23we don't have, you know,
53:24great indices of of of
53:25that from our sample because
53:26they really don't know.
53:27So I think what we
53:28need to do is move
53:29to getting people to bring
53:30it in and, you know,
53:33start to take a you
53:34know, start to actually analyze
53:35the constituents of what they're
53:36actually using.
53:37And then a lot of
53:38them are using, like, CBD
53:39products and all of that
53:40too. And I know it's
53:41typically seen as a very
53:42mundane thing, but I'm quite
53:43interested in trying to quantify,
53:45you know, some of that
53:45as as well.
53:47And then the other option,
53:48obviously, in some things, it
53:50just it's hard from a,
53:50you know, a biomarker in
53:52blood samples or or in
53:53urine and the like. There's
53:54just not great index indices
53:56of these with regard to,
53:57you know I think we
53:58could look at highly frequent
53:59users and get, you know,
54:00really matched people according to
54:01that and then use some
54:02of those as indices of
54:04kind of amount or frequency
54:05of of exposure. But,
54:07yeah, that's one of the
54:08big limitations also of of
54:10these, obviously, that I wish
54:12you could get better of,
54:12especially from an endocannabinoid
54:14system, perspective.
54:16And I'm quite curious personally
54:17about CB two and microglia
54:19and the like and inflammation.
54:20And I feel like it's
54:21a very understudied
54:22area.
54:24There is one study I'm
54:25working on with, Andrea Edlow
54:27at Harvard as well as
54:28a few other people where
54:29we are,
54:30getting placental samples, and then
54:32her lab is developing them
54:33into microglia,
54:34individualized microglia. It's across it's
54:36more to opioid, but we're
54:38gonna have a fair number
54:39of cannabis users in that
54:40too. It's from the HPCB
54:41study, so a kind of
54:42a sub study. So I'm
54:43hopeful that some of those
54:44approaches may be useful as
54:45well. And then,
54:46any studies that are collecting
54:48the delivery biospecimens,
54:50you know, maybe getting some
54:52better, indices of this through
54:53the umbilical cord tissue.
54:56But, yeah, I think it's
54:57a a big limitation.
55:00Other questions?
55:02I see a question in
55:03the chat from Uri. Do
55:05you wanna unmute and ask
55:06your question?
55:10Yes. I can do that.
55:12First of all, thank you
55:13very much for your talk.
55:14It was really, really interesting.
55:17Maybe you said it. I'm
55:18not sure. You said it
55:20about the last,
55:21slide you showed, but in
55:22general, did you
55:24look for differences
55:26between edibles and smoking?
55:28Like,
55:30did you even have enough,
55:32data about
55:33edible use? Yeah. So in
55:34a b c d, no.
55:36Just because it was all
55:37retrospective report from ten years
55:39ago.
55:40These kids were recruited in
55:41twenty sixteen. So I would
55:43just like to keep in
55:44mind that it means they
55:44were being born around twenty,
55:47two thousand and six.
55:49So I think there are
55:50some things in that dataset
55:51that would suggest there are
55:52likely less edibles around at
55:53that time period,
55:55for them relative to their
55:56ubiquity now.
55:57So I think that was
55:58probably more traditional use.
56:01But in our more our
56:02our studies we're doing here,
56:03we're getting we're really intensely
56:05quantifying how they're using it.
56:07And most everybody
56:08it's I wanna say it's
56:09ninety five percent. I know
56:10it's greater than ninety. I
56:11wanna say it's ninety five
56:12percent now, though, are mostly
56:14smoking it with very little
56:15edible use. But, again, I
56:16think that's a a difference
56:17with our under resourced sample
56:19that is largely getting these,
56:20from the black market still.
56:22Yeah. But it makes sense
56:23if it was from two
56:24thousand and six. So Yeah.
56:26Yeah. Probably. So and another
56:28question is, did you have
56:30data about
56:32postnatal exposure, about the passive
56:34smoking?
56:35Yeah. This is a great,
56:36great question.
56:38We don't have a good
56:39data on that in ABCD.
56:41We are getting those data
56:43in,
56:44the CUDL dataset, so I'm
56:45really curious to look at
56:46that. And what I wish
56:47we could have gotten that
56:48we weren't able to get
56:49would be, breast milk.
56:50I think that'd be very
56:51interesting. In our sample, it
56:53gets a little conflicting because
56:55a lot of them are
56:55not breastfeeding.
56:56A lot of the our
56:57individuals across both groups are
56:59not breastfeeding. So a lot
57:00of them are using formula.
57:01So partly because of that,
57:02we decided not to look
57:03at that in our budgetary
57:04restrictions. But I think,
57:06you know,
57:07especially given the the fattiness
57:09of of of breast milk
57:10and stuff, I think the
57:11idea of cannabis consumers getting
57:12in there and binding and
57:13transmitting is is is quite
57:15quite, like, quite strong.
57:17And I I do wanna
57:17emphasize that, yeah, I think
57:18that's very important because even,
57:19like, as little as, like,
57:21five ish years ago, I
57:22wanna say, the,
57:23American College of
57:25Obstetrics and the like was
57:27actually saying they couldn't recommend
57:28not using or using cannabis
57:30during pregnancy or during breast
57:31milk feeding, and they subsequently
57:32reversed that saying that, you
57:34know, there's not enough data,
57:35but they would not recommend
57:36it now. But they were
57:36just saying they couldn't say
57:38yes or no.
57:39So I think, you know,
57:40given some of these things
57:41are I think we just
57:42need a lot more research
57:43in this this area, unfortunately,
57:45before we can really make
57:45a lot of policy decisions.
57:47But
57:48Alright. Thank you very much.
57:50Oh, well, thank you. And,
57:52just for anybody, I'm happy
57:53to if you wanna get
57:54in touch by email or
57:55anything like that. I know
57:55we're about out of time
57:56now, but I'm I would
57:57be happy to to continue
57:58these conversations offline too.
58:00Thanks
58:01for
58:02a incredible presentation, and I'm
58:04I I will be getting
58:05in touch with you. I
58:06have some interesting ideas that
58:07we might wanna think about.
58:09Well, Cyril, I've been reading
58:10your work for decades, and
58:11I love your work. So
58:12I'd love to collaborate as
58:12I'm going to. Thank you.
58:15Alright. Thanks everyone for joining.
58:16We'll see you next month.