Scientist Stories: Olivia Corradin, The Genetics of Opioid Addiction

Good afternoon, everyone. Welcome to today's webinar, The Genetics of Opioid Addiction: What We Know, What We Are Learning. I'm Rick Young, member of Whitehead Institute and professor of biology at MIT, and I'll be your host and moderator today. Today's webinar is part of a series of talks that we've organized to share Whitehead Institute's science and research with the wider community. In a few moments, we'll hear from our presenter, the newest Whitehead Institute member, Olivia Koretyn. First, I'd like to let you know that this webinar is being recorded for future viewing. Also, we will be fielding questions at the end of today's presentation, but you can submit a question at any time using the Q&A box at the bottom of your screen. With that, I have the pleasure of introducing our speaker today. Olivia Corridan. Olivia obtained her Bachelor of Science in Biochemistry from Marquette University and her PhD in Genetics and Genome Sciences at Case Western Reserve University. Her thesis there focused on the study of genetic and epigenetic dysregulation of gene expression in human disease. This work earned her the Doctor of Excellence Award in Genetics from Case. Olivia joined Whitehead Institute as a fellow in July of 2016. And was named the Scott Cook and Signe Osby Fellow. In 2017, she was awarded an NIH Pioneer Award and an NIH National Institute for Drug Abuse grant to support her studies of substance abuse disorders. This year was an exciting one for us at Whitehead as Olivia became a member of the Whitehead and an assistant professor of biology at MIT. A typical human has about 2,000 structural variants in her genetic code. In other words, each of us has differences in our genetic code that affect about 1 in 1,000 base pairs. Here at Whitehead, Olivia uses pioneering approaches to study how these genetic variants are associated with human disease. The opioid epidemic is a pressing public health crisis. Opioid overdoses have become the leading cause of accidental death. This crisis has been exacerbated during the COVID-19 pandemic. Genetics is a major contributor to opioid use disorder, and understanding the genes and gene variations involved may well reveal critical aspects of the biology of addiction. Today, Olivia will discuss with us her pioneering studies on the genetics of addiction and will describe the changes she's identified in the brain and how these changes converge on genes critical to the disorder. Thank you for presenting today, Olivia.

Great, thank you so much, Rick, for that introduction, and to all of you for coming today. I'm excited to share with you some of the work that my lab has been doing, which really focuses on using human genetic approaches to help us understand the underlying basis to opioid addiction. So long before the COVID-19 pandemic, we were already dealing with a major public health crisis in the United States, and that's the opioid overdose epidemic. Back in 2017, opioid overdose deaths became the leading cause of accidental death, exceeding that even of automobile accidents. And this problem has only been exacerbated during the COVID-19 pandemic with reports of a 25 to 40% spike in overdose deaths in the last year. So understanding this public health crisis is going to be really important for the treatment and help of many people in the US. Our goal here today is to talk a little bit about how we got to the state of the opioid epidemic and how we might think about using genetics to address this problem. So you may or may not be surprised to learn that opioid use disorder is estimated to be about 60 to 80% heritable. So today we're gonna discuss how we've arrived at that number, what we know about the genetics of substance abuse disorders, and I'll share a little bit about our work which looks directly at postmortem brain tissue samples from individuals who died of an opioid overdose. So before we can discuss the genetics of the disorder, we have to think a little bit about how this epidemic came to be. I find it quite striking to learn that 80% of people who use heroin first used a prescription opioid. This really calls into play the role of prescription opioids in the crisis that we are dealing with. And this is really an epidemic fueled by misinformation. Perhaps the best example of this is a letter that was written to the editor of New England Journal of Medicine that was published back in 1980. And this letter concluded that addiction is very rare amongst individuals prescribed with an opioid. Here is that letter in its entirety. This entire study was only about 4 or 5 sentences, involved looking at hospital records of individuals who were given a single dose of a narcotic during their stay in a hospital, and then looked for documented cases of opioid addiction in this group. And they concluded that there was a very rare connection between opioid prescription and addiction. The problem is that this small study was then cited by many pharmaceutical companies to help convince doctors that opioids were safe and should be used widespread. What makes the opioid epidemic particularly challenging is that opioids do remain one of our best approaches for treating chronic pain. It's important to keep in mind that there is a positive use for opioid prescriptions. The World Health Organization, at the same time that this letter came out, declared being free from pain as a basic human right. And it was perceived that back in the 1980s, we were undertreating chronic pain. The problem is that this study was not directly looking at chronic pain, but rather a specific instance of observed opioid use. So we have to think about what that might mean for our studies of addiction and understanding its prevalence in the US. So this study released here led to an explosion in the prescription of opioid overdoses, perhaps best shown by this particular plot. Here we're plotting all of the opioid overdose deaths that occur each year, being attributed to pharmaceutical-related opioids in yellow, heroin in orange, and synthetic opioids like fentanyl in red. You can see as that yellow bar started to increase in the early 2000s, and this is very viewed as the first wave of the opioid epidemic. As people began to realize the negative consequences of having readily available opioid prescriptions, and these began to be more heavily regulated, we see a turn to a second wave of the epidemic, which has been prescribed as an increase in heroin-related opioid overdoses. As patients unfortunately turned to street drugs to deal with their chronic pain. This is followed by the third wave of the epidemic, which we are currently in, which is driven by the use of synthetic opioids like fentanyl. The challenge here is that these synthetic opioids are much more potent, and that makes it far more easy to overdose on the substance. In order to see the extent of the effect of this on the opioid, we have to actually zoom out to see how much this has grown in the recent years. So as you can see, there's a lot of complex factors that play into this epidemic involving government, medical doctors, true chronic pain issues, and regulation. But what we want to discuss is beyond just the environmental component. What are the biological components to opioid addiction, and can we use our understanding of genetics to make an impact on such an important question? So the first question you may be asking yourself is, how do we even know that there is a genetic component to a disease like opioid use disorder? And the way in which we can ask this question is to take advantage of a very naturally occurring experiment in the human population, and that's twins. Naturally occurring, a phenomenon, two individuals being born with the same genome. The way we use twins to look at genetic studies is to— genetic disorders is to compare monozygotic or identical twins to dizygotic or fraternal twins. We look at twin pairs who are raised in the same household and assume that they have pretty similar environments. We then can compare identical twins to fraternal twins and look to see what is the relative genetic contribution to a particular disorder. So when we apply this kind of strategy to substance abuse disorders, what do we find? This is where we get that estimate of opioid use disorder being about 60 to 80% heritable. And that number can be hard to put in context, so for comparison's sake, height is estimated to be about 80 to 90% heritable, schizophrenia is around 80%, and multiple sclerosis around 70%. So this is really on par with other complex human traits. You can see opioid use over here on this plot. So similar heritability levels between opioid use and cocaine use disorder. We see slightly lower heritability as we go down to the left end of the scale. We have to be careful here. This is the metric of populations, not a metric for individuals. So on a population level, we see about 60% of the disorder being something that we might be able to prescribe to particular genetic variants. But how do we go about identifying those genetic variants? Well, the most common approach used to look for the genetic contributors to these disorders is a genome-wide association study. And on the surface, these studies are relatively straightforward. We start by collecting a large group of people that are afflicted with a disease and a large group of controls and then assay their genetic material. We're looking for DNA variants or single nucleotide changes called SNPs and looking to see if any of these individual SNPs are found at greater prevalence amongst the cases than in the controls. So we identify a genetic variant that's had higher prevalence in our cases versus controls. We say that the variant is associated with increased risk for disease. So these aren't necessarily causal variants. They're not necessarily causing the trait to occur, but what they are doing is increasing one's likelihood of developing disease.. And so, we can use these as basically genetic clues to understand what biological mechanisms are most important to the pathology of the disease that we're studying. So, over the last about 10 years or so, these approaches have been applied to a wide variety of diseases, including substance abuse disorders. So, what have we found so far? So, substance abuse disorders have found a number of key genes related to their heritability. So, for example, alcohol use disorder, there's two key genes that have been identified. Both evolve directly in alcohol metabolism. The number one locus associated with nicotine dependence is involving a cluster of genes that encode nicotine receptors. And for opioid use disorder, the top associated locus involves the gene OPRM1, which is an opioid receptor. So, at the very top of our association results, we find that the genetic variants linked to the trait tend to be directly related to the substance that we're discussing, and this raises questions of whether or not there's any genetic component or shared pathophysiology between substance abuse disorders. In other words, is there an underlying phenotype of addiction? We can better ask these questions by not focusing on individual DNA variants but looking at genetics more genome-wide. So we can take a strategy which is called genetic correlation, which leverages the fact that in these studies we're actually looking at millions of DNA changes. So we can ask for a given genetic variant, if it has a strong effect on trait 1, does it also have a strong effect on trait 2? And we can do that for the millions of DNA variants that we've studied and look for traits that have a significant correlation. Correlation with one another. You can also use the strategy to look at traits that have opposite effects or a negative correlation. When you apply the strategy to those substance abuse disorders, now we find that substance use disorders really have a significant genetic correlation. So beyond the top hits of these diseases, there seems to be a lot of the underlying genetic architecture that contributes to multiple different substance abuse We look specifically at opioid use disorder, we find that it correlates with other substance use disorders like alcohol use disorder. But it also genetically correlates with other diseases that are more neuropsychiatric like depression or schizophrenia, as well as behavioral traits, such as being tolerant of high-risk behaviors, having a tendency to take risks, which is reflected in number of sexual partners and even your likelihood of speeding, while driving a car. And so this gives us a nice global picture of how some of these traits may relate and gives us a way to dive deeper into the study of these disorders and certainly validates the conclusion that substance use disorders are a disease in and of themselves. So, from these studies, we often see hundreds of genetic variants linked to many human traits. Including some of the traits related to alcohol drinking behavior and smoking behaviors. When it comes to the actual study of opioid use disorder, relatively few genetic loci have actually been substantially associated, the most famous being the opioid receptor gene that I mentioned earlier. So why, if the disease is heritable, haven't we been able to identify the genetic contributors to that heritability? So one of the major reasons for this is simply study power. In order to identify these genetic associations, we need a lot of cases and genetic information so that we can see the effect even though everyone lives very different lives and has different habits. In order to see the effect of genetics, we need to amass really large sample sizes. Because opioid use disorder is historically underfunded, this study lags behind many of the complex traits that we study in the field of genetics. We also have to consider if this trait may be different in some way that would affect the way we design genetic studies. So we have to ask ourselves, is the study design actually appropriate for the underlying biology of the opioid use disorder trait? So we have to take a step back and think about the trait itself. So I think of opioid addiction as a chronic multi-stage disorder. It's characterized by multiple stages that often cycle. And so we see these stages including exposure or first-time use of a substance, followed by a period of binging or intoxication, often followed by symptoms of withdrawal, which can either relate to recovery or relapse. And each of these stages has its own set of genetic and phenotypic variation. So for example, just think about the factors that influence whether or not one will be exposed to an opioid. This has been shown to be linked to one's general impulsivity, willingness to take risks. Think about people who have to have surgeries following a major car accident or following other adventures that might be related to their likelihood of developing an exposure. It may also be related to one's pain tolerance, the relative dosage, and the amount of time that you spend taking the drug itself. And that's just in determining who will be exposed. After exposure, we also see variation in response to the drug itself. You can see this at a population level because 8 to 10% of people exposed to opioids will end up developing opioid use disorder, and 20% of heroin users will end up developing opioid use disorder. But even when you speak with patients who suffer from opioid use disorder, they often describe their first experience with an opioid in a very particular way. They describe the subsequent cycles of the disorder as trying to recreate that first experience. And so understanding just the phenotype of exposure is critical to our understanding of why some people develop addiction and why others do not. Another area of major variation is in withdrawal. We see that there's a wide variety of severity of withdrawal symptoms, and this likely contributes to the variation in relapse rate, which is about 60 to 80% for opioid use disorder. And this is really a critical component to the disorder because it's in relapse where we see the majority of overdose deaths. It's at the end stage of the disorder where patients will return to a dosage that they are no longer desensitized for. And so it's really important to understand what aspects are going to contribute to not only who develops opioid use disorder, but who will go on to overdose. And so you can see how designing a study is going to be very particular to which phenotypes we can capture. So if we go back to our general paradigm for a GWAS study, we're looking at individuals with opioid use disorder compared to individuals without. What we don't know is whether some— any of the individuals without opioid use disorder were never exposed to the substance in the first place. So some of our control samples may have genetic vulnerability to addiction, but that never manifests into the disorder because they were simply never exposed. So it's going to identify a very particular part of the disorder. Recently, groups have been directly asking this question by collecting samples of individuals who are exposed to an opioid but do not go on to develop opioid use disorder. Now you're asking specifically from post-exposure to developing OUD, What are the genetic variants that contribute to this phenotype? We must also consider that not all individuals with opioid use disorder are the same. There's going to be variation here, and understanding who with opioid use disorder is at most risk for overdose is also going to be an important part of this disease. And how we design our studies is going to dictate what we can learn from these approaches. So, so far we've been talking about directly looking at DNA sequence and assessing how genetic variants may attribute to the trait. But there are many functions to the genome and many ways that we can measure it. So for example, we can look directly at transcription or RNA levels in patients to see how the genome is functioning in terms of producing transcripts. These transcripts then, of course, go on to become proteins, and those are other measures of genome function that we can assess to look at how the genome is working without even directly looking at DNA sequence. Now there's one more critical element that tells us about how your genome functions, and that's the regulatory elements. So only about 2% of your DNA actually encodes proteins. The majority of it is playing a role in the structure of genome and the function and regulation of gene expression. And gene regulatory elements, or enhancers, are really critical to this process. These elements will define what genes are turned on and what context. It's what allows your neurons to be different from your liver. And so measuring the function of these elements is also going to be a way to look at how the genome is working. And we really care about these elements because many of the genetic variants associated with disease fall within these regulatory elements rather than directly within the protein coding region. And so instead of looking directly at DNA sequence, another strategy to understand genetic components to disease is to directly assay patient tissue samples. This gives us a chance to more deeply phenotype the samples as well as capture both genetic and environmental alterations that may occur during the course of the disease. So for example, if you might hypothesize that opioid exposure could change gene expression, that's something we can capture by directly looking at different patient tissue samples. And that's the type of study I'm going to describe to you today. So I've been involved with a multi-team collaboration involving researchers at Mount Sinai, RTI, NSU, and Case Western. And we have worked to look at postmortem brain tissue samples of individuals who died of an opioid overdose in the hopes of furthering our understanding of what makes the brain respond differently to opioid versus control samples. So in this cohort, we've collected 51 different opioid overdose samples. These all have confirmed opioid overdose as a cause of death and have a medical history of substance use disorders. Our healthy controls are all accidental deaths, so these mostly involve car accidents. They have a negative toxicology and no history of substance abuse. And we went through these samples and collected neurons from the particular region of the brain called the dorsolateral prefrontal cortex, which is really critical in decision-making. And the assay that we performed to help us characterize how these two samples differ is a technique called ChIP-seq, and this allows us to measure the function of regulatory elements in the genome. So here's an example of what this data looks like. We have these little hills here that represent regions of enrichment for the protein that we're interested in with our ChIP-seq assay and reflect how much activity we see at that particular regulatory element. The gray is showing you a snapshot of some of our control samples and the blue some of our opioid cases. And what we're looking for is differences that distinguish these regulatory element activity from either the cases or controls. All together, we have about 600,000 different regions of the genome that we're interested in. We're trying to characterize where do we see the most distinct changes in the genome. And so we applied a really well-established approach to look at the global differences that distinguish two groups. And so using the strategy, we identified about 388 specific peaks that had reduced or hypoacetylation in the opioid cases, so evidence of reduced activity. What we didn't find were any massive changes that significantly altered between opioid overdose and controls. So even though we're looking at confirmed cases of opioid overdose, we don't see large extreme changes even at genes involving the opioid receptor, for example. What this suggests is that there's a lot more heterogeneity or subtlety in the effects of opioid use on the brain. So to give you a sense of all of the differences we did identify, what we observe is about a 30% reduction in signal between our cases and our controls, so really suggesting that the effects we observe are relatively modest, which is interesting in and of itself. And so we're interested in some of the target genes of these subtle changes and what they might mean about the pathology of the disorder. One of our most interesting targets is this gene DUSK4, which is involved in a well-known pathway called MAPK signaling. But this also raises some other questions about the nature of the changes that we have in our samples. And so I set out to ask the following question. Are there any regulatory elements that are strongly dysregulated, but only in a subset of our cases? Perhaps the phenotype and the response to the drug is so heterogeneous that by looking for things that are common across all of our samples, we're missing some of the interesting sites. So our goal was to identify regulatory elements that distinguish an individual case from our full set of healthy control tissues. So here there's a regulatory element shown in gray, these little hills, for multiple of our control samples. And the traditional approach would be to look for regions where we see a subtle but consistent difference across our cases compared to controls. Our goal here was to identify things that are more specific to individual samples. So for example, We look for regions where there is a reduction in activity in a specific case, even if it's not found in all of the cases. And we go one case at a time to define these differences. So we call this approach variable enhancer loci mapping, or VELS for short. And so we went through and identified the VELS, or the regions that make individual cases distinct from our panel of controls. We identified a total of about 80,000 different peaks that distinguished individuals from the controls. My very first question was, which of these vowel peaks are shared in more than one of our opioid cases? And so we looked to see how often are the vowels found in multiple cases. What I found is about 10% of our vowels were found in two or more opioid cases. So the vast majority of what we had identified were really individual-specific. But one thing we have to keep in mind when we study regulatory elements is that these elements tend to work in a combinatorial fashion. So we don't see a one-to-one relationship between regulatory elements and genes. Instead, there are frequently multiple regulatory elements that all work to control the same gene. So instead of focusing on just the peaks themselves, we turned our attention to the genes that are regulated by each of these regulatory element peaks and assigned each of these to their target gene. Here we asked, how often is the target gene found across multiple opioid cases? And here we saw a lot more sharing. So while only 10% of the vowels were shared across multiple samples, 60% of the target genes were targeted by vowels in multiple samples. So what this suggests is that the loss of enhancer activity that we were observing is not randomly occurring across the genome, but rather occurring in very specific patterns and perhaps converging on specific loci. And so we sought to go after this particular question: where is the most convergence of these changes occurring in the genome? So we worked with our collaborator Richard Solari from Axial Incorporated to identify a strategy to quantify this convergence. And so we looked at this problem from multiple levels. We look for regulatory elements that are all linked to the same target gene. And we first asked, well, are there any samples where there is multiple regulatory elements that are lost or dysregulated in the same patient? So one form of convergence would be if one person lost several elements regulating the same gene. Another form of convergence would be if the same element was lost in multiple samples. And then the third, more complex form of convergence is when elements that target the same gene are lost in distinct samples. And we consider all types of convergence when we identify those loci with the greatest co-occurrence of variant enhancer sites across opioid cases. Using the strategy, we identified 5 key genes that had really substantial convergence of these Vel events or this loss of regulatory element activity. And this included some genes like the potassium channel gene and a GABA receptor gene that really agree with other studies of opioid addiction, particularly in model organisms. Our top locus involved the gene ASTN2, and I'd like to show you just how much convergence we're observing at this locus. So here each of these elements are shown to be linked to the gene target ASTN2. Of all 106 elements linked to this gene, 69 of them had a Val in at least one of our opioid cases. So now what you're looking at is each row represents one patient, and we've just put a darker dot here if it's a Val in that sample. And what we found is that 11 of our 51 cases had at least one Val targeting this gene. You may notice that there are some samples where we see a large number of hypoacetylation events or loss of regulatory element activity occurring within the same patient. And so we're really seeing a multi-level convergence at the regulatory element gene sites at this gene. And this pattern can vary from gene to gene. So one of my other favorite examples is this gene DUSK4, which is involved in the MAP kinase pathway. Here we see the same element is lost in in multiple cases. But then we see a lot of convergence on the diagonal. So there are really many elements controlling this gene. Almost half of our samples have at least one val targeting this gene. And so this is one of those genes that we do identify by both strategies, looking globally, using a— looking at single peaks at a time, and also by using this convergence strategy. So both have complementary advantages. So finally, we wanted to go forward and think more about the 5 loci that we had found this VEL convergence for. And to do this, we use a technique called heritability partitioning, which sounds pretty fancy, but it's really just an approach to look at a fraction of the genome at a time. So we care about the particular fraction of the genome where we see our VEL events occurring and converging. And we ask, do the genetic variants in those regions contribute to one's likelihood of developing any other disease that we've studied. And when we looked at this, we found some interesting results. So we looked at a total of 25 different disorders and asked, do the regions that have VEL convergence also explain any of the heritability for other diseases? We found 3 diseases in particular. This includes generalized anxiety disorder, educational attainment, and number of sexual partners. Now, both educational attainment and number of sexual partners have just shown to be genetically correlated with opioid use disorder, and all three traits have been linked epidemiologically to opioid use disorder. What's interesting is these three traits are also strongly linked to early life adversity, or in other words, high-stress events in childhood. So this does call into question whether some of the variants that we're looking at, some of the vowel changes that we observe, may exist prior to opioid exposure. We're capturing the end stage of a disease, and it may be capturing multiple causal roles for the disease itself. But overall, a really strong result suggesting that our variant enhancer loci conversions are really connected to phenotypes related to opioid use disorder. And pointing out some really critical target genes. So finally, we compare these results to the latest opioid use disorder genetic study. We ranked our genes based on just how much about convergence we had, so that ASTN gene is at the very top of our rank. And then we compared it to the top SNPs or genetic variants linked to opioid use disorder so far. We find is a really striking pattern The genes with the most devel convergence tend to occur near the top SNPs associated so far with opioid use disorder. Particular interests are multiple genes involved in GABA signaling, potassium channels, and MAP kinase signaling. So altogether today I showed you how we can look at postmortem brain tissue samples to identify those changes that may be influenced by multiple stages of the disorder. And what we found is that these epigenetic or chromatin changes are co-occurring with regions that have genetic vulnerability to traits like generalized anxiety disorder, number of sexual partners, and educational attainment. These also tend to occur near our current best estimates of the SNPs involved in opioid use disorder. What we hope to do is utilize these orthogonal types of to help study how different aspects of opioid use disorder might manifest. By looking at this complex disorder from multiple angles, we hope to reveal those aspects of the disease which play an incredibly critical role. Ultimately, we'd like to use this information to help inform ways in which we can develop preventative measures for opioid use disorder as well as treat the disorder directly. So with that, I'd like to thank the members of my lab for their efforts on this work. The members of the Opioid Epigenetics Consortia that is involved in the study that I described today are listed here. Of course, our funding sources and you for your time and attention. Thank you.

Thank you so much, Olivia. If, uh, just a reminder, everyone can put their questions into the Q&A box. But I'd like to start just by asking you, how generalizable is your approach using VELS for the broader spectrum of diseases that are genetically associated?

That's a great question. Yeah, so we've certainly thought about applying this to other strategies that have looked at postmortem brain tissue and even in cancers. And so there's, you know, we hypothesize there's likely to occur in multiple traits, but we certainly haven't done those studies yet. So we think there's some good potential there.

So I'm going to read you some questions from your audience. There's a question about the spikes in patients with opioid use disorder during COVID-19. Have they occurred because of return to use or as a consequence of fentanyl being cut into these substances?

Yeah, I think it's a pretty complex answer is likely to be what's going on here. And there's still gonna be additional work done to really nail down exactly what's contributing. Certainly, one of the factors is that people had a harder time accessing treatment during the COVID-19 pandemic. So things like methadone clinics, which require daily access to medical care, that could be a contributor, as well as increased rates of depression. These are really quite complex relationships. So I think that there was a spike beyond the growth that's expected from fentanyl use. And the exact contributors to that remain to be seen.

Is there a difference between men and women with respect to who's more genetically predisposed to opioid use disorder?

There's some differences in rates of opioid addiction in men and women. Certainly, we see a greater prevalence in males. Whether that's genetically driven directly is unsure. We still have enough sample size to divide our population and see what the genetic factors may be that define that difference.

Yeah, men have a higher risk behavior than women generally across the spectrum of risky behaviors.

Sure, there's also concerns that women might be undertreated for pain. And so they're certainly less likely to receive an opioid prescription, which would in turn lead to reduced rates of addiction. That's not necessarily a good thing in this scenario. Treating pain is obviously still an important problem.

Could you say a little bit more about how SNPs factor into the vowel convergence?

Yeah. So we're really talking about two different types of studies, one that looks directly at patient tissue samples and we identified functional changes in the genome. And then we simply compared those results to other studies that looked at genetic variants. The next step for us is going to be looking at the genetic sequence of patients in our cohort to see if they have any genetic variants that contribute to those changes that we observed in the brain tissue.

What are your own thoughts about treating opioid use disorder with other prescription drugs?

I mean, it certainly has been effective for treating overdose itself. If we can think of ways that may counteract some of the aspects of addiction itself, that's certainly worth thinking about. I think there's a huge opportunity, though, for prevention. Understanding who's at greatest risk, the mechanisms by which we observe and recontact patients that have a prescription, regulation are all on the table here as well. And genetics has a role there. If we can understand genetic factors that put you at risk that might influence, you know, simply how many pills you're prescribed at a time, or require seeing a counselor while taking the opioid. And so you can think about really practical ways to prevent the disorder if we understand who's at greatest risk.

Do you plan to do parallel animal model studies?

Yes, there is some really interesting model organism work that do behavioral assays in rodents to actually measure different aspects of this behavior. It's quite a specialized set of skills, so that's something we'll do with a collaborator perhaps. But the work that they're able to do to investigate the target genes we identify is quite elegant. It's just very difficult. So our goal is to get the very best target genes worth studying so that we may work with individuals who do those types of studies.

There's some curiosity about whether you'd look for genetic correlates to increased withdrawal symptoms and/or relapse. And if, if so, could you describe that study design?

That's the, you know, that is, you know, the holy grail of understanding overdose would be to characterize relapse and withdrawal. But that is a particularly challenging study to undertake. You would need to be working with a clinic that's treating opioid use disorder. And so we'd certainly be interested in doing that, but to this point, I'm not sure anyone is currently enrolling people in a study like that.

Here's a question that we've discussed before. Given the convergence of VELS to specific genes, could you identify the same genes if you simply compare gene expression differences rather than histone-modified ChIP?

Yeah, it's a great question. So it's asking, could we just assay the RNA directly and identify the same effects? I think it's a little bit challenging here, because it's practical. So looking at chromatin, postmortem tissue, DNA-protein interactions are a little bit more stable. And so we get more consistent results. Than you do when you're looking at RNA levels directly following death. And then the other aspect here is that, you know, RNA levels are a snapshot in time, whereas the chromatin state may be reflecting a more prolonged or consistent effect on cellular function. So we've looked at the correlation between the changes we observe in the histones versus gene expression, and we certainly see you see a nice correlation between loss of regulatory element and decrease in gene expression, but it's not so strong that you could identify the genes without looking at the histones themselves.

While recognizing how difficult it is to obtain 2 times 51 sets of postmortem brains, it's still a small set. Do you have plans to enlarge that sample?

Yeah, it is one of the, you know, more difficult parts is that this is a very large ChIP-seq study, but overall a pretty small study in general compared to other genetics studies. So yeah, we're interested in expanding, but I think first we have to make sure that we're focused on the right brain region. So one of the things we'll go after first is asking just how different is the chromatin structure in different regions of the brain. And consider another brain region, the nucleus accumbens, when we're looking at these differences and then go after the best bet and expand studies for that particular brain region.

How might these genetic risks ultimately be utilized by medical professionals when deciding upon pain treatment for surgeries or other medical procedures? Will it help modify practices?

Yeah, there's a lot of effort to generate what's called a polygenic risk score. It's a way of calculating one's likelihood of developing disease by using a combination of genetic changes instead of just looking at one at a time. And that has the potential to at least give patients more information. Whether medical doctors will use that to choose how to dose or prescribe an opioid is not necessarily the only question. They may also choose to simply spend more time educating the patient, give them more resources if you knew that you're at greater risk. This is analogous to families that have a history of alcoholism. By simply being aware that you're at greater risk of developing the disorder and putting off the use of alcohol in childhood before the age of 21 can also lower your rate of alcoholism in your family. So thinking about practical ways to reduce risk for those most at genetic risk is where I've been thinking about it the most.

Have you observed any relationships between pain tolerance and opioid use disorder?

So it's a great question. Actually, one of the top genes that we looked at in our study is this gene called ASTN2. And at first, we didn't know much about this gene. It's not been very well studied. But recently, there was a genetic study that looked at pain tolerance following surgery, and this gene was one of the associated loci. So they were looking at patients' interest or demand for opioids following surgery. And so we think there may be generally a relationship between pain tolerance and opioid addiction, but the relationship has a lot to be determined yet.

Are there implications for how the genetic work could contribute to prevention, particularly among children or adolescents? For example, education, anxiety issues, and even sexual activity make great sense as targets for prevention.

Yeah, so it's, it's the causation here is, uh, is still under, um, under, understudied. So whether there is, it's the high stress event early in life that can contributes jointly to both traits. So early life adversity can relate to anxiety disorder, which then directly relates to opioid use disorder, or it may independently contribute to both traits. So understanding the effects of stress on the brain and stress signaling, and whether that's related to what we're observing in opioid addiction, would all be interesting questions to ask.

Why did you decide to investigate postmortem brain tissue? Could you have obtained genetic information from living individuals to do a similar analysis?

Yeah, so one of the advantages of looking at postmortem tissue is that we're able, first of all, to study the neurons specifically, focus on the part of the brain that— and the genes that are involved in the function of the brain in response to opioids. When you're looking at genetic material, you don't necessarily know what the consequences what tissues being affected. So it's one advantage. Another is that we can obtain just a little bit more information when, in terms of defining our cases and controls, so get more detailed toxicology report. We know that an individual will never develop opioid use disorder because we're looking at them at the, at the end stage of the disease. In particular, Understanding opioid overdose itself and whether that's different from OUD, opioid use disorder, is of major interest. And so this gives us a chance to directly study the end stage of the disease and that part that we're most interested in preventing.

What portion of the population doesn't respond to opioids as painkillers?

Oh, that I'm not familiar with. Yeah, so there's definitely a range of response, but I'm not familiar with any of the studies that may have looked at those outliers.

Are there any plans to expand the collaboration to other medical research sites to access additional samples?

Yes, so we certainly be interested in incorporating as many groups as possible. We're looking at collaborating with gene expression studies and of course the genetic studies. But so I think the only way to tackle a disease like this is to get as much collaboration as possible.

Do you see these kinds of genetic screens being used in the future with fetuses similar to other types of genetic testing that occur now?

You think, you know, you can dream of a future like that where you're able to understand your risk every time you take any sort of medication. There are risk factors involved, and being more informed of what you in particular are at risk for, I think, would be incredibly useful in terms of informed consent and informed patient decisions.

How do you review comorbidity with psychiatric illnesses and underlying trauma? For example, do you get medical history with the postmortem tissue, or is that a review of existing data?

So with our samples, this is done with our collaborator Deborah Mash, and she looks at both medical history and she talks with families. So we're able to obtain information through there. So we know our samples in particular are not comorbid for other psychiatric disorders. And so we try to focus the study as much as possible on opioid abuse.

Have you confirmed changes in protein expression levels for the genes you have identified as correlated with addiction?

Great question. No, we have not. We've gone as far as the RNA levels. Kind of a different way of preparing the sample and storing the sample if you want to look at protein. And so there may be some opportunities for us to look particular at some protein levels in, in some of the samples that we have, but the studies that we have been done so far are from what we call flash frozen tissue, which is best for these chromatin and RNA experiments.

Can you say anything about the roles of duration of disease, say emerging versus chronic, and severity of disease or tolerance play in your research?

You know, for us, you know, those are, you know, those are excellent questions. We certainly would, would be relevant. Unfortunately, just getting that amount of information is what's challenging, and we certainly don't have enough patients to stratify by those types of phenotypes. And so we're certainly dealing with the goal of trying to get as large of a study as possible while not including too many confounders. And so it would be nice to look at those variables, but we're just underpowered to do so thus far.

I thought it was interesting you described a confounding problem when selecting controls for OUD. There may be false negatives, for example, people who have a risk variant but never experience opioid use. Are there other ways to find these individuals and, and identify the false negatives?

Yeah, so, you know, what we— I think the field is moving a lot towards the use of these large biobanks which collect genetic information as well as a large number of phenotypic information, everything from how much coffee you drink to MRI images of your brain. And so now that we're starting to collect all of this information, if we have the entire history of one's experience with opioid prescriptions, you could look at those individuals that have had a lot of exposure to opioids but did not go on to develop substance abuse, studying that population itself might be really telling, sort of the inverse of addiction, those who are resistant. And so I think there will be opportunities for studies like that as these large medical biobanks start to create resources available to researchers.

Have you seen any dorsal root ganglia gene expression differences in OUD?

We have not specifically looked for that, so I can't let you know.

Maybe. Well, Olivia, that is the end of the questions, and I just want to Thank you once again for such a wonderful, clear presentation. And to the audience, thank you all for joining us. And Olivia, once again, thanks for sharing this really interesting work.

Great. Thank you, everyone.