Video Investigating the Environmental Etiology of Autism Spectrum Disorder Play Pause Volume Quality 720P 720P 576P Fullscreen Captions Transcript Chapters Slides Investigating the Environmental Etiology of Autism Spectrum Disorder Overview Magdalena Janecka, PhD, discusses the environmental etiology of autism spectrum disorder. So I want to thank Magna forgiving the first talk of the fall season of this um, psych lecture series as some of you know, it was started kind of in Covid first to present some of the covid work, but then also as a forum for people in psychiatry to learn about all the research in psychiatry. And I think it was quite successful and there were enough people who said they'd like to see it go on. Even that we've started to get in the fall, it'll be less frequent. Um, but it will have the same, um, you know, star cast of president presenters from the department if you'd like to present. Uh, let us know, um, let me know, original And uh, I think it's for me it's a great pleasure that magnet is giving it a kick off for the fall season for the 2021, 2022 seasons. Magda received her PhD in psychiatric genetics from the Institute of psychiatry in King's College. She went on to Oxford to do a postdoc. And then, um, she was recruited by ivy right Lindbergh to join his team as a senior postdoctoral fellow. And I think we are extraordinarily fortunate that we were able to convince her to turned out competing offers in the UK and stay at Mount Sinai and I wish it became an assistant professor at 29 2019. And in spite of all the obstacles. Yeah, all of us experience the obstacles, you know, coming from abroad and all that. She managed to build a very, very successful team, has really enhanced the epidemiology uh and related work in the super center as well as the Mount Sinai. So um thank you for agreeing to present and you're talking and watch it. No unclipped. But people have questions, you know, just to use the just use the chat function or sorry, the Q. And a function. And I will help moderate that as needed. Thank you Magda. Thanks joseph. Thanks for the very kind introduction and thanks everyone for Janine. I just realized yesterday actually that this will be the first time for me to present to the department's faculty. So of course it's a great honor to be here and I'm really looking forward to showing some of the work done in the lab of you today. Um and what my lab has focused on since the inception about two years ago is studying whether environmental causes of autism exists. Um and if yes, what are they? Hopefully again, I won't lose the entire audience if I admit that we largely don't know at this point, it turned out we need a bit longer than two years for that, but I'm hoping to show you why it's both really important and really difficult to get this right. Mhm. Huh. Mhm. Um So I'll start from making a case for the importance of better understanding the role of the environment in autism despite consistently very high estimate ability heritability. Ah then I'll discuss some of the challenges in getting things right in this field of research and how my lab has been tackling these questions for the last two years. And as I go along I'll of course acknowledge the team members living specific projects. But since these guys are such a great team I wanted to start from giving a quick shot down to all of them. Um and so to give you a very brief background about what we know about the etiology of autism. Thanks to some seminal studies by my karate, we've known for decades now about the prominent role of the genetic component autism despite at the time no insights about the specifics of quote what is inherited and over the time we've learned a little bit more about these inherited factors as well as the genetic causes of ASD that arised in over So skipping over decades of discovery really by 2021 many specific common and we're genetic variants have been identified for the efforts of large scale international consortia And expanding the sample size from this 21 person. Fallston and battery study to millions of individuals linked through a range of complex familiar relationships were declared by spend standing from the silvery center Estimated the asd heritability to be about 80 adding to the firm body of evidence suggesting that S. D. S. Of predominantly genetic origins which exposes a huge elephant in the room really and makes us pose a question whether why do we even bother studying the environment in the context of ideology of autism and whether all the thousands of studies exploring are either wrong or tapped into a very negligible component of what is relevant and in part to answer these questions, I think it's helpful to spend a bit of time discussing what the heritability measures practices actually tell us. So in brief heritability estimate, which is this 80% in the context of autism informs us about the variants in the phenotype that is attributable to genetic variation in a given context. So, a lot to unpack, Which is an incredibly useful metric, but one key caveat in inferring that genetic factors alone explain 80% of autism in theology is the non independence between genetic and environmental effects on the linotype. And the first issue here is that correlation between genetic and environmental variation or in other words, that extend towards some of the environmental exposures are genetically influenced. So taking lung cancer as an example, lung cancer is always an example. We know now it can be caused by smoking, but we also know that propensity to smoke itself is under genetic influences, which already gives us a feeling that deciding whether it's shade is genetic or environmental, is going to be difficult. And in heritability studies a common assumption is that shade variants is independently influenced by variation in genetic and environmental factories. But in the presence of gene environment correlation to estimate this effect correctly, one needs to account for this gene environment correlation. Otherwise, if genes and environment are positively correlated, the correlation component will inflate the variance attributable to the genetic effects. And secondly, a similar inflation of heritability estimates may happen due to gene environment interaction, which happens when certain genetic variants make us more susceptible to the effects of the environment. So, under this situation again, simplifying things a bit, neither a genetic or environmental risk factor on their own are sufficient to result in disease and the disease only occurs if both the genetic and environmental factors act together and similarly as before, if the environmental factor is tend to run in families, the variance explained by them will increase the estimated genetic effects in standard heritability models. So the heritability measures are useful to the extent they provide the upper band for the effects of genetic variation on the phenotype. But it's important to remember that they may also include the effects of gene environment correlation and interaction and in SDI research we mostly still don't know how this genetic component exerts its effects and deserve the risk, including potentially through the environment. And just to be very clear here. These considerations reflect a discussion in the field that really has been going on for decades. Um and to anyone who would like to find out more. I really recommend the spectrum article by brian lee as well as the very comprehensive review by Tennyson and Harry. But to keep the argument balance. The other side of this going is that the association between many of the environmental factors Nasd might occur only through background genetic confounding. So going back to this diagram, disregarding for a second, everything else we know, it's plausible that small case are more likely to get land cancer because smoking causes lung cancer, which here is represented by the purple arrow, or because the genetic variation underlying smoking propensity also increases the risk of lung cancer, which here is represented by the navy blue arrows. But with no causal effect in that blue arrow situation between smoking itself and the disease. And it's very important to be able to distinguish between these two scenarios because only in the smoking causal situation, increasing the rates of smoking in the population would result in lower prevalence of lung cancer and similar causal structure can be related to the etiology of autism with for example, some prenatal environmental exposure and autism as an outcome. And if I was to summarize my research with a single graph, it's probably this with all of us trying to distinguish between the purple and navy blue pathways. Um, and the quickly illustrate this point further with one of his current projects utilizing the spark example, we've seen higher rates of exposures like brief hypoxia or preterm birth in Children with ASD compared to their unaffected siblings, which replicates the association between these factors and autism that have been widely widely reported in the past. However, an interesting pattern emerge wants to stratify these numbers by simplex versus multiplex family status indicating respectively, families with their single or multiple members affected with fasd. And what we saw there was for both hypoxia and preterm birth was that these exposures were considerably more common in multiplex families. And while we've only recently began to explore what these patterns may mean, The simplest explanation for higher rates in multiplex families is that these families have higher genetic liability for autism, which accounts for the multiple members affected with the disorder and that there is a play a tropic effects of this genetic variation on both a sc and the exposure. And here the prevalence of preterm birth in multiplex controls equally. In the one observed in simplex cases suggest that perhaps there are no causal effects of these exposures themselves on autism. Of course, like nothing ever is this simple. Um, and these factories do need to be investigated further, but it's definitely a possibility we need to consider on the other hand, what emerged from various analysis was that there were certain exposures with prevalence is higher in ASD cases compared to their unaffected siblings. But with the rate in the simplex families on par with those observed in multiplex families suggesting that familial genetic liability to asD may be unrelated to the risk of these exposures and the association between these exposures and autism would not be driven by genetic play, atrophy alone. These are still very early results and we're still checking if these patterns were not just something artifact actual. But the only point I want to make for now is that the relationship between our genes, environment and authors, um risks are complex or maybe complex and we probably shouldn't either hide or discard things too early. Well again, these things can be very complex and we may not yet know what this 80% heritability estimate means. I just want to be clear that it's difficult to dispute that there is a number subset of individuals with ASD whose diagnosis attributable to the genetic causes alone. And it's fair to say that if someone has a pathogenic mutation in one of the key synaptic scaffolding proteins, likely any additional contribution of the environment is going to be miniscule or zero. And secondly, the fact that this heritability estimates may underestimate the role of the environment. It doesn't mean that they do. And as I mentioned earlier, the 80% heritability of autism Whereden the upper band of this effect, but it's also possible that 80% is what it is. But leaving this complexity aside for just a very brief moment, environmental research on SB is a really big field and as illustrated here by a graph from the meta analysis by dR model Bernini also from Sinai highlighting the range of environmental exposures associated with ASD in previous studies. The fact is significant in the meta analysis included a bunch of things. Prenatal exposure to accessorize birth, trauma, maternal gestational diabetes, obesity, cesarean delivery and importantly over and over again. And perhaps I think now it's time to say it again, vaccines were not to any extent associated with autism risk. But as you see this, really long as you see, um there are these really long paris along the Y axis, indicating that there remains a huge amount of uncertainty with regard to the strength of the association between these exposures and SD, which may reflect variable exposure definitions across the studies and our estimates obtained from relatively small samples. So, for example, for birth trauma, this effect is anywhere between 50% higher risk to 16-fold higher, risk. which are quite different numbers. Um, if you ask me. But except for these statistical or measurement issues, as I have signaled before, many of the epidemiological studies do not account for the potential genetic confounding. I've touched upon in one of the earlier slides. So, for example, an association between maternal medication use Nasd Can be observed even if the true causal association between two is absent. But instead, as in classical confounding scenario, their shared genetic factors play a tropically affecting both the both the disease risk for which the medication is taken and SD and the second caveat here is that the environment is more complex and we do not really get to study these exposures like material max, cesarean delivery in isolation from everything else. Um, and to illustrate the point about this complexity of focus on Ss arise, which is a class of antidepressant medications that have been consistently linked with an increased risk of autism, leading to a debate of the role of potential confounding by indication i. E. Whether the observed associations are not predominantly driven by the effects of maternal depression or even its underlying genetics. So, ours and other groups have done some work in that direction, suggesting that this confining is likely the case. But to complicate things even further, it also seems that depression itself is not an isolated risk factor. So this graph, it's taken from another project that varicocele durian has done in the lab and it's not also available in my the archive. And it illustrates quite well that women with a mental health disorder in pregnancy which included mostly women with depression and anxiety, have higher rates of a series of physical conditions, including digestive circulatory and and the current conditions which really highlights the possibility that the effects of depression or its underlying genetic factors on autism risk may be observed separately to the correlated effects of these additional physical comorbidities in mandatory in pregnancy. And this likely applies to a wide range of environmental exposures that we often think of and try to study in isolation. Some exposures will occur in clusters were not always aware of. And while it's difficult to always account for everything analytically. Our interpretations of the observational associations should always acknowledge this complexity. And so I think it will take really intense efforts to move beyond this observational associations to a real understanding of why I was seeing this effect. But I do think that it's something that is worth our while, at the very least, to conclusively show that some of these exposures do not cause autism and lessen potential fear around them. Secondly, if all we ever get to, we'll be showing that all environmental associations we see arose due to background confronting alone, knowing about the play a tropic effects of some genetic factors on both. For example, depression or preterm birth, N. I. S. D. It's still important in understanding the etiology of autism. So, having justified myself for the last um 15 minutes or so for the remainder of the talk, I'll focus on several studies done by the lab members. So this will be mostly short vignettes. To highlight some of the ongoing projects, but I'll be happy to elaborate on the details later on during the Q. And a. And just as a short primer, our labs focuses on early life risk factors for autism, particularly those that can affect the fetus before birth, in line with the hourly neuro developmental origins of autism. And while we're interested in the association's themselves, our focus really is on distinguishing the potentially causal associations from those that arise to various sources of background confounding, including by factors like parental genetic, medical or demographic factories. So the first study I'll highlight focuses on the role of maternal health in pregnancy and etiology of autism. And while several associations between maternal diagnosis and offspring have been suggested, including maternal depression or asthma, lack of any systematic studies, means that majority of maternal diagnosis have never been looked at in the context of etiology of ASD. And also for the previously reported associations, we mostly do not know to what extent they are driven by genetic and funding or other maternal comorbidities. And addressing these knowledge gaps is important to do. To better understand the etiology of ASD. To at least give ourselves a chance to identify potential preventive strategies and to improve the design of other epidemiological studies by understanding which maternal health factors could act as potential confounding res. And this project gave rise to the first funded arrogant application in the lab, where our aims are to identify maternal diagnosis in pregnancy associated with the risk of autism in offspring. Using both maternal medical records from danish register and Kaiser permanente northern California, cooperative and secondly, to interrogate the level to which these associations are driven by genetic confounding by triangulating both family and genetic data. So, in a pilot study, to get preliminary data for this proposal, we ran this diagnosis wide association study between maternal conditions in pregnancy and offspring risk of autism and this graph represents the significant levels of individual I. C. D diagnosis which are color coded by their broader diagnostic categories. So the higher the dot the more significant association, the great dot at the bottom indicates diagnosis that we're not significant after rigorous filtering process including adjustment for confront others, multiple testing, correction and permutation testing and diagnosed that remained significant after all these steps were then analyzed in a joint model in order to account for the comparability between them. And we ended up observing that some genital urinary, metabolic digestive and psychiatric diagnosis were independently associated with an increased risk of autism in offspring. What was more curious though so to say was that there was also a range of maternal diagnosis associated with lower risk of offspring ASd. And this list mostly included either mild diagnosis or medical codes indicating contact with a health care provider. And so it seems that likely what is driving these patterns is that women with higher healthcare utilization and therefore more likely to receive relatively mild diagnosis are also less likely to have a child with ASd. We're still thinking of some explanations as to why this could be the case, including lower subclinical mental health problems in women with higher healthcare utilization but we still haven't really produced an explanation we feel satisfied with. But the key message here I think is that this association should not be ever taken at face value and I think there is something interesting going on here and definitely something we need to think more about. But it's definitely not that tonsillitis or dermatitis caused autism. And since the Israeli study under again, start post at Kachadourian sleet. We've refined our approach a bit and moved to a much larger dany sample. So the preliminary findings from there to an extent replicate those in Israel with some differences in the specific diagnosis associated obviously, but falling into the same broad categories with the exception of the genital urinary ones, which were not significant after comparability adjustments. Two additional clusters of diagnosis that emerged included injuries and birth complications. And similarly, as in Israel, we observed this pattern of mild diagnosis and codes indicative of contact with healthcare associated with lower risk of piracy. After this stage, we separated non chronic conditions and explored the evidence for the underlying genetic confronting is in comparisons between exposed and unexposed sibling which allowed us to control for family level can founders and also we compare the effects of maternal diagnosis occurring in pregnancy to those to those that are more temporarily distal from pregnancy. These two analysis countries be done for the chronic conditions because the timing of the diagnosis in the register reflects when the woman saw the doctor for her condition rather than the disease onset. And because this event seeing a doctor in this season said may really be years apart, would really have very low confidence in deciding which it was exposed and which wasn't and one relative to pregnancy, that this is really started. Um, but for the results first for the sibling comparisons, um what we found using a stratified cox regression was that there was no evidence for family level confronting, And similarly, these non chronic conditions when diagnosed 2-4 years before pregnancy were mostly not associated with autism. Risk, again, indicating lack of evidence for family welcome funding. So while we haven't identified any evidence for genetic confounding this association so far, um we're really at the very start of this project. So things may still turn around drastically, and over the next four years will be working to replicate these results in a Californian sample to incorporate additional designs for investigating the genetic and non genetic effects underlying these associations, which will include paternal and cousin comparisons, as well as apologetic risk approaches. And we if we still at that point c um, that there is no evidence for genetic and funding, There will there's plenty to be done in terms of studying where the non genetic signal is coming from. So here by design were not able to distinguish between factors like direct diagnosis related effects, maternal genetic variation that is not transmitted to offspring, but it can influence still some other s irrelevant factories, maternal use of medication and likely many other factories. Uh, and the second quick vignette, I wanted to show some of the is the work, sorry, is the work on the role of maternal exposure to medication in autism. So we're definitely not the first ones to think about this with multiple studies having already shown that several classes of medications taken by pregnant women which included antidepressant accessorize, anti epileptics and asthma medicines are associated with an increased risk of I. S. D. In her offspring. And if the medication is in pregnancy is causally related to asd, it's of course critical that we know it. But the problem in this field is all the background confronting going on, including by maternal indication. And as I already mentioned, um it's really difficult to get around it. But um like some of the reasons why it's important to get this right is um first of all, some medications maybe mechanistic lee associated with ASD. And secondly, because right now many women may stop taking the medication they need only because it was shown to be associated with autism due to background confounding. And so I think there's a lot of responsibility and how we do and how we communicate this, this research to others and our approach to study uh this effect. Why, Sorry? While reducing the confounding by the underlying diagnosis is to create these groups of medications based on their pharmacological targets. So as you see here, Jax A to e are prescribed for different conditions as indicated by differently colored dots, but they all exert their effects for acting on the same target. And once we get this target based group of medications composed of medications prescribed for diverse conditions. We can test whether maternal exposure to its underlying checks is associated with offspring, dynasty and to the extent that we do get such indication heterogeneous group. This approach reduces the confounding by indication, by loosening the tight link between the drug use and condition present in traditional approaches. It also allows us to increase power to identify shoe signal because there will be more Children exposed to the target group compared to the individual medication. And finally by teasing out the mechanistic signal. It can also highlight the biological systems that are sensitive to prenatal disruption by maternal medications. This signal is also easier to replicate in other cohorts because the medications underlying each group may differ without any consequences on the influence about the mechanisms. And this is a project that we've really been working on for a number of years in collaboration with the Department of pharmacological Sciences, which really has been one of the most fruitful and fun collaborations I've experienced and who is spearheading this analysis has some really exciting updates, which she will get to find out about soon. Uh And in the meantime if you're attending the ICMP this year, I really encourage you to check out his post today. So maybe a bit of suspense in terms of the result, the latest results but I do promise it will be worth it. Um And with the next Vignette, I wanted to illustrate how studying the environment in autism might help us better understand. Not only what are the factors associated with the disorder as in binary status of either having ASD or not but also those factors that contribute to its differential presentation in different individuals. And this time I'll actually show you some part of the press results from the recent project using this for example. So here we focus on additional comorbidities of I. S. D. Which just as a quick background are known to be highly prevalent and really affect all body systems including epilepsy, gastrointestinal symptoms or A. D. H. D. And what's important here is that different individuals would present with a different constellation of this. And what we still don't fully understand is what drives the association of these conditions with autism and why different individuals can present so differently from each other. And part of the rationale for the design of this study which is kind of being finalized by far where the observations from syndrome IQ forms of autism caused by for example, mutations in a DNP or shank three genes and both of these mutations are associated with both SD and a constellation of other medical issues. And so what we wanted to ask was whether in a similar fashion environmental factors associated with ASD are also associated with different constellation of additional comorbidities. Furthermore, we wanted to test whether there is any evidence that these comorbidities arises a downstream consequence of autism or are causally independent of it. And before I move any further, nothing we did here implies that we believe that any of these associations between the environmental factors and autism vulnerabilities is truly causal. So in other conclusions, we account for the fact that what may really be going on is that these exposures simply tag underlying causal genetic factories. So to start unpacking all of this. Slowly, first, we did observe that all of the environmental exposures that we wanted to study occurred at a higher rate in Children with autism compared to their unaffected siblings, Illustrating 1st, the association between these factors and autism potentially confounded by the background genetics. So we've got this part of the model already. And next we observed that among autism cases these different environmental risk factors were associated with different comorbidities. And what struck us here was that among the strongest associations we observed many of them were already establishing the general population, including, for example, association between early brain trauma and epilepsy and antenatal infection and deafness, a lead poisoning and obesity or fetal alcohol syndrome with microcephaly. And this really underscores that the mechanisms underlying comorbidities in autism may not be qualitatively different from what we see in unaffected individuals. And to explore this idea, we looked at the association between the same exposures and comparability is in unaffected siblings and by doing so, testing whether the association between the environmental factor and this cummerbund conditions, I somehow mediated by autism and here because of the sample size for the siblings in the spark study, the sample size. Sorry, there were many of you are unaffected siblings compared to autism cases in the SPARK study. Therefore, we focus on the two most common exposures including preterm birth and very hypoxia. And what we expect for both of them was that while the point estimates were often actually higher in siblings, the confidence intervals around this effect were very much overlapping cases and controls, suggesting that there is no evidence that the association between these environmental factors and comorbidities has anything to do with having or not having autism. Um, and so for the interim summary, it does seem like certain environmental exposures are associated with both SD and it's comfortable conditions with no evidence that ASD is a mediating factor. And for now, the way we interpret this result is that first autism and these comorbidities share some of their causes and secondly, higher prevalence of these comorbidities in autism may simply be a dosage effect. The case is having more of the underlying costs. Big environmental or genetic. Um, and now for the next penultimate lynette, I want to change gears a bit and give you a quick primary on some of the epigenetic work happening in the lab. So here the rationale behind this work is that to some degrees, both genetic and environmental factors may influence the risk of autism may and because they both affect the epigenetic looking at things at this level, we may be able to capture both of these sources of effect and focusing on DNA methylation Similarly as the genetic contribution to autism. We expect some effects terrorized due to both common and where or private events. And for the common side of things, we've got a well established classical heroes approach which is largely analogous to what she was. And the Manhattan plot I'm showing over here um is coming from the latest published ste was study illustrating quite well that we're probably still underpowered to detect common epigenetic signal in a heterogeneous disorder like autism. For their epigenetic variation methods are much more recent and there's still no gold standard in the field really. But Andrew Sharpstein advance Sinai has done amazing work at advancing this field and the results shown in the next couple of slides relied strongly under work described by purpose. It'll, so when it comes to this where DNA methylation outliers or this private, very private events. Most of these events are only seen in one or two individuals not allowing us to assess the penetrations association with asd of any of these individual events. However, what we can do instead is to a simple burden test checking whether these events are more common in any of the diagnosis which are which were interested in which in our study included autism. A. D. H. D. Comorbidities, the S. D. A. D. H. D. Etcetera. And we run this analysis in two independent Danish array based samples. Isaac Minerva and Isaac both of which profiled DNA methylation is in neonatal black spots. And the results from these two samples were subsequently meta analyzed together. And this work has shown here was supported by a great collaboration with the Isaac consortium as well as my nerves have grant from the Dprk and we observed slightly elevated burden of this DNA methylation of pliers in both A. S. D. And A. D. H. D. But only in the cummerbund cases. This effect was statistically significant and this really links well with observations of higher clinical severity and genetic burden uh In this from orbit individuals we also observed in both samples an increased load of this very epigenetic events and individuals with intellectual disability. And given that intellectual disability is enriched for these rare genetic mutations. These results time with work done by Michael Brynn and Sylvia, the rubies from the civil center on epigenetic disruption in some carriers of rare genetic variants. Um Next given our interest in the early environment, we tested whether certain early life environmental factors are associated with higher burden of this rare epigenetic variants. And hopefully by now it's clear that when I say environmental what I refer to is measured on the environmental level, acknowledging the possible background genetic influences and among these environmental exposures. We observed an association between the burden of this rarity, DNA methylation variants and measures of prematurity both measured in weeks and grams and inferred from the DNA methylation data. So there is definitely some interesting signal to follow up on. And right now we've got another super talented post doc in the team. Doctor Christine Hanson on the task. And over the coming months she'll be using the sample of mono and di psychotic twins. We're both this and concurred them for mental illness and family shows to detect this very epigenetic variants and further interrogate the heritability of these events. To strengthen the evidence for their potential pathogenesis, pathogenesis City, assess the recurrence across different psychiatrist disorder and last but very much not least she will contribute to the methods development in the field. So what space and look out for Christine's work and in fact you might do so very soon if you're attending either W. C. P. G. Or S AJ or christian will be contributing an oral and poster presentations and last but very much not least. Um I just wanted to give you a quick primer on what our team is has been busy with since joining the lab literally a few weeks ago which is trying to better understand the potential causal role of preterm birth in autism. And while the association has been extensively reported including by the recent study by our wonderful colleagues and sand in from the city center it's possible that despite careful controlling for the background confounding including for example maternal health or smoking in pregnancy. Some residual confounding remains and of course we can never be sure that the reported associations are truly caused out. But seeing that the results from several different methods all points in the same direction, we may start feeling a bit more confident about what we're observing. And one such additional alternative approach that we can add to this battery of tests is mandatory in Random Ization which simplifying things a bit or a lot allows us to test whether individuals who genetically are the higher risk of an exposure are also at the higher risk of the outcome. And what our team is currently working on is meta analyzed in the existing gestational eyes GeoS data to identify genetic low side that are robustly associated with gestational age which will serve as genetic instrument in this model and in the next stage. Um this will also allow her to test whether the association between pre and post and birth and autism occurs for the same more distinct mechanisms. And finally, in the near future our goal will be to interrogate whether changes in DNA methylation which we know is strongly associated with the station. All age mediate any of the effects of gestational age on either autism or other outcomes associated with pre or post and brief. So for the last time today, uh the look out for the updates by the team members. And uh yeah, I think that's uh that's in terms of the results and in conclusion, I hope to have managed to confide that the environment is worth studying and if all we ever achieve is proving its lack of involvement in office and the theology. I consider that an effort that is very, very well. The high heritability estimates may indeed reflect the fact that autism is predominantly genetic in origins but given the impact of gene environment correlations and interactions, I think it's good to keep an open mind about this for now at least and again, there are some autism cases with fully genetic etiology. We just don't know yet what fraction of all asd cases they really are confounding is pervasive but with a mix of observational genetic and epigenetic data, I think we can slowly work our way through this and some of the examples of how which I approach this in the lab include projects on maternal health and medications in pregnancy for now trying to distill respectively, familiar and clinical confronting and importantly, even if these things are still a bit murky. Um even without fully understanding the causal relationships between different factors studying environment in autism can still help us better understand the underlying architecture of risk and deserve their heterogeneity and last but not least integrating epigenetic and mandarin, random ization insights may help us better understand the functional aspect linking these exposures to neurodevelopmental and neuropsychiatric illness and the lab is currently doing both applied and methods development work in this direction and with that I'd like to thank all of the amazing collaborators who contributed to different projects. That's very part of the stock, both within and outside and outside. Sinai first and foremost, thanks to my amazing team and Avi. And joe from the Silver Center, Avi for mentoring me for the last 10 years since I started my PhD and Joseph for his helping joining and navigating early years, his faculty here. Uh, and I'd like to thank many of the wonderful faculty menders here at Sinai, my founders and all of you for listening. And I'm ready to take questions if there are any, like backed up great change the view. So it's great talk and to remind people of questions should go into Q and A. Um, maybe I'll kick it off. So when I try and diss genetics, which I do sometimes, but rarely, especially heritability estimates, you know, the heritability of height is even higher than how to parent really wants it to, but height has changed systematically over decades. Right. And clearly there's an environmental component of some measure. And the reason it's not measured is because there's a huge cohort, in fact, right? When people look at a certain cohort, and so obviously the diagnosis of autism has changed dramatically, even more dramatically. I would say that height. So I guess I should know this. Like how in the sanding paper, how was that dealt with it, then, is there a way to incorporate that? Even with Gillian randomization. Right? Because the diagnosis, I'll shut up. Yeah, I think, yeah, some fire center is that I don't know really whether that whether and how that's handled in Spence paper and something at all. But I think that exposes one of the key things about heritability estimates that maybe are not discussed often enough. Heritability estimates assume, and if they are very specific to a given context. So if, for example, we were to completely equalize everyone's environment, Then every all the variance you would see would be completely genetic heritability would be 100 if we were everybody smokes. Exactly, Exactly. That's that's so the way we react to this environment is determined by genetics and that's what we would see in heritability estimates. Um, if on the other hand, we were to give, let's say force all women to undergo, let's say assuming things that are causal undergo, for example, birth complications or take some terracotta genic medication in pregnancy. Same. We would see higher effect of genes over environment. Um, so I think, I think both with between, like within a certain period of time between different different cohorts as well as within the same, let's say side over the time we will see these estimates change, which exposes like art, a factual thing about arte factual, however we call it about heritability estimates that they are not something that's like immutable characteristic of the disease itself of art sites as you kind of give an example for height and I think maybe like it's often not spoken enough. I think it's very, I think it comes up sometimes in discussions about educational attainment and trying to estimate heritability of education or socioeconomic status. You know, the heritability may be lower because schooling is universal, but once we take away universal schooling heritability estimate would go up because we would see, um, this effect variants to measure the, it's a short answer. Yeah. I guess I guess, you know what I often say in the circumstances Because of all the reasons you said, you know, I honestly don't know if it's worthwhile. I can whether heritability of certain disorder or traded 60, 50, 70 or I want to know if it's non-0. Yeah, exactly. Yeah. Yeah. Yeah. Yeah. No, I used to like in the center where I did my PhD like estimating estimating heritability was one of the things many people did for many years. And uh, I remember one of the academic being like, Oh, it's always between 40 and 80. Like what does it matter? It's good to know that it's non-0. No, I think like, no, I think I don't want to be like, I don't, I'm not a little critical about the measure. I think a very good measure. Once we understand exactly the caveats around it, I don't see how people are. Maybe they're using the chat. People use the Q. And A. I will also check the chat, but please. Uh, I know you have questions out there. Oh, so, um, happy to have open discussions about anything. Hmm. So I'm one thing that maybe you could expand upon is obviously ePA genetics and environment and genetics, you know, they obviously correct. I'm sorry that someone intersect right, but particularly the idea that environment is somehow reflected in the genome. I know there's been a long term interest of yours and like where are you going with that? Where are we going with that? I think for now. Well, we're, as I highlighted, there are two main avenues were trying to take. One is for christians project looking at the very events and to what extent they may actually contribute to uh etiology of asd to ask better understanding impact of even like understanding the impact of genetic influences on on risk. Uh and secondly, I think Project Artemis will move onto which is looking in integrating DNA methylation measures with dandelion random randomization concept framework to better understand to artifact, to what extent these things can actually mediate. Um, the effects of exposure on an outcome. I think we're still very limited by the sample sizes that are available. That's the critical thing and you know, for what's available. Often these samples are not very well characterized. So I think like at least for now a lot of work done in this fear remains preliminary. Yeah. So there's a related question from Michael about DNA methylation. Can you elaborate on some of your analytical coaches study epa genetics and autism. My predictions about epigenetic age or genotype based predictions of DNA methylation which you kind of just touched upon but take it away. Yeah, I'm trying to unpack my first questions. Um epigenetic age and then kind of editing methylation que TLS which I guess would roll into your Yeah, I think both are kind of widely studied, widely studied in the context of psychiatry. These are not really like things were incorporated into our analysis. Um things we often account for. Um But yeah, we don't really explicitly study these things and everything on paternal. About the other thing you talk about a and they want to know about The complement of eight. So everyone should talk about paternal. I spent my PhD studying Katrina and she knows what I think that's Yeah. Um so because men do not carry pregnancies, I think there is like one critical kind of period of exposure is out of the equation. Having said that, you know, the whole field of better understanding the effects of environmental on sperm on offspring is still very active. I think because of how many kind of connections we need to make. I don't think we're at the point where we can say anything any of the paternal exposures, our causal of offspring outcomes through the effects of the spiral. But that's of course like a subject of active investigation. And yeah, often often were the way we're using that as an is as a negative control for exposure in pregnancy. That maybe given the effects on the spring. That may be too simplistic. Um but yeah, something something we're actively thinking about. Mhm. And of course, paternal age, um which my PhD was on and have it paid for the mice. My technologies are probably the most well established risk for environmental, it's complex, but risk factor for autism. So maybe you have to go back to Michael's question. So, if you're thinking about, you know, if you have it's a good kind of chihuahuas for autism. Yeah, yeah. Kind of methylation. Q Tl in your and Dylan radicalization. Yeah. And I think that's how they are actually used in the framework that that's what's kind of you need for the method to work. You need the background kind of genetic laws. I underlined the um, CPG Lohse you're interested in. So the MQ T m sorry, I missed that. MQ tl are actively like part of the framework. I can I can forward the relevant paper, Michael later on. Okay, well, um thank you for a great talk and uh happy fall to everybody. And Happy because one of the main New Year's has started recently, Maggie, thanks for a great talk and and thanks everyone launching the this this this series again. Thanks everyone Published November 30, 2021 Created by Featured Faculty Magdalena Janecka, PhD Assistant Professor of PsychiatryMount Sinai Health System View full profile
